This study explored time trends of inequality of distribution in healthcare resources at municipal level and district level in Shanghai from 2010 to 2016. It confirmed the reduction of the inequalities of medical resources distribution in Shanghai after 2009’s new reform. Additionally, we investigated the changes of inequality of doctor allocation at the municipal level and health institution allocation at the district level, and explored spatial changing patterns and geographic clustering features across 16 districts during these seven years using the spatial autocorrelation method.
Firstly, we discovered that the inequality of healthcare resource allocation was largely improved after reform. However, some types of healthcare resources were unevenly distributed at the municipal level (i.e., doctors), or at the district level (i.e., institutions), and these disparities persisted or even became more severe afterwards. Theil indices of indicators were improved both at the municipal level and district level during the seven years, indicating that Shanghai government’s targeted policies, especially the “5 + 3 + 1” program targeting at rural areas, have greatly reduced the inequality in healthcare resource allocation. However, the inequalities of some types of healthcare resources still preserved, taking inequality in doctor density at municipal level and institution density at rural level for examples. It reminded us to cautiously evaluate the degree of policy impacts. Though the targeted policies were purposely designed for the rural areas to reduce the inequality in healthcare resource allocation, the problems of maldistribution of physicians had been unsolved due to the long training cycle for health workforce and other social factors that contribute to the workforce maldistribution. For health institutions in rural districts, it was harder to attract and retain health workforce due to the existing condition of lower salary and socio-economic constrains. Hence, though opening or expanding branch institutions and facilities in rural areas were easy to achieve, workforce shortage has limited the volume and quality of services. Unfortunately, the inequality in institution density in rural areas was worsen by expanding the health facilities, possibly because the policy’s preference to build new institutions was often given to the prioritized rural areas that were relatively developed compared to the underdeveloped rural areas, which resulted in an unfair distribution of institutions among rural areas. Some similar studies were in line with these findings[37–38]. Therefore, more detailed area-specific healthcare resource planning and allocation policies are required to balance the health workforce and health facility distribution for Shanghai in the future.
Secondly, we further used the spatial autocorrelation analysis to reveal the spatial changing patterns of healthcare resource distribution across different districts and identified priority areas in Shanghai from 2010 to 2016. To our best knowledge, Theil index method could measure the gaps among all the units, while spatial autocorrelation analysis identify how these gaps are distributed on the map. In this study, using the spatial autocorrelation, we illustrated different spatial concentration status of the distribution of five types of healthcare resources across Shanghai from 2010 to 2016. Through global spatial autocorrelation analysis, it was found that strengthening of spatial concentration effect in healthcare resource of beds and workforce in Shanghai over time. This distribution pattern was also proved by the results with the local spatial autocorrelation analysis that the high-high type clusters were all located in central administrative districts and low-low type clusters were all located in rural ones for beds and workforce resources during the seven-year period. These findings were unsurprising, due to the fact that economic strength and geographic position play vital roles in influencing healthcare resource distribution. Compared to other regions, the central urban districts with high economic strength were unlikely to suffer beds and health workforce shortage and are better positioned to cope with the shortage of these healthcare resources. To solve the problem of misdistribution of healthcare resources, Shanghai’s health reform had helped disadvantaged districts to improve their resource shortage with various incentives. For example, the “cold spots” districts of institution density including Songjiang, Jinshan and Fengxian (displayed as low-low cluster areas in Fig. 3) were improved by the expanding of branch institutions and no longer being the “cold spots” in 2016. Similarly, such low-low clustering of bed density in Jiading district, workforce density in Chongming district disappeared gradually. However, some urban areas showing high-high clustering and rural areas showing low-low clustering of health resource density were co-existed from the beginning to the end of the study. This indicated that healthcare resource concentration tended to be correlated and agglomerated together. The aforementioned findings further added evidence on the existence of geographic imbalances in healthcare resource distribution in Shanghai. In the context of regional economic integration, theoretically there should be spatial correlations between the healthcare resource concentrations among different provinces and cities[39]. As a mega-city, Shanghai’s urban areas are unexceptionally more attractive to health workforce due to the advanced economic, cultural, educational and professional advantages. The suburban areas are less economically developed, and residents face longer geographic distance to health facilities and poorer public transportation, all lead to less access to healthcare service, and fewer career opportunities for health professionals. To solve this problem, besides efforts from the suburban areas to develop the economy and living conditions, multi-sectoral alliance from related governmental sectors and mutual collaborations between develop and undeveloped regions are also needed. The finding was also proven by some prior studies [5, 6, 38, 39].
Thirdly, this study also provided evidence on identifying the most-in-need-of-help areas in healthcare resource re-distribution by further detecting the priority areas that should be firstly considered to improve the equality of healthcare resources in Shanghai. In this study, the high-high cluster units for beds and workforce distribution can be found in JA, HP and XH district, and the low-low cluster units can be found in southwest and south part of Shanghai, especially QP, SJ, JS and FX district (Fig. 3). It also indicated that disparity of beds and workforce still kept unchanged in these areas after 2009’s health reform. These findings provided more solid evidence for area-specific workforce policy making in Shanghai, aiming at improving the beds and health workforce in rural low-low and low-high cluster areas. For example, some point-to-point cross-regional medical pairing assistance programs should be suggested to re-allocate healthcare resource between districts with shortages of beds and workforce (JA, HP and XH), and districts with abundant healthcare resources (QP, SJ, JS and FX). This pairing programs will make more accuracies by paring up the geographical regions who are most in need and who are comparatively most sufficient in certain subtypes of health workforce. In addition, some eHealth technology can also be applied to fill the workforce gaps in rural districts in Shanghai [41].
The present study has some limitations that should be acknowledged. Firstly, we measured healthcare resource allocation with five quantitative indicators without considering their quality aspects. Therefore, quality indicators measuring healthcare resource distribution should be developed in the future to complement this study. Secondly, for data availability, the study used municipal-level and district-level data from the officially published yearbooks to measure the inequality of healthcare resource from 2010 to 2016, which might cause the problems of the modifiable areal unit problem (MAUP) [40]. Data at county-level or community-level would be more ideal to sketch the geographic distribution of healthcare resource in future study. Moreover, a future study on following up thte changes in healthcare resource allocation should be carried out when the newer data from 2017 to the present are available.