Measures of transportation accessibility for ECs under different transport modes.
Referring to the accessibility-based measurements proposed in Hulland and Van Wee 40, this paper reckoned cumulative opportunities for visiting FCs based on the travel time of ECs 41. Stemming from 2016 urban resident trip surveys in Shenzhen, more residents prefer traveling in non-driving models, accounting for 74%. But it is safer and faster to drive during the outbreak of the epidemic. Therefore, taking 15minutes, 30minutes, and 45minutes as a measurement index, we calculated the cumulative opportunities of visiting FCs under the two models.
Figure 2a, b shows that within 15 minutes, 77.45% of the ECs traveling in non-driving mode could not arrive on time to the nearest FCs, while only 3.43% in driving mode could not reach. Most of these communities are located near district boundaries, such as the west of Longgang District, the east of the Yantian District, and the south of Longhua District. Figure 2b, e shows traveling in driving mode within 15 minutes is equivalent to traveling in non-driving mode within 45 minutes. Similarly, the number of inaccessible ECs and FCs is very close between the two models, accounting for 3.43% and 1.47%, respectively; Still, there are 45 communities accounting for 22.06% that cannot reach the nearest hospital in the non-driving mode while people there can achieve it by driving within 30-minutes travel thresholds(Fig. 2c, d).
On the whole, traveling in non-driving mode means more travel time, resulting in some people turning to seek treatment in non-government designated medical institutions. This way of treatment is not conducive to timely diagnosis and follow-up for patients, thereby increasing the risk of prevention and control.
Rationality evaluation of medical resources.
According to the plan of the National Health Commission of the PRC, by 2020, China will achieve a 30-minutes primary medical service circle. Bosanac 42 advocated a similar criterion, so this paper investigated the cumulative opportunities of visiting FCs under the non-driving mode in ECs within 30 minutes, to explore the balance between medical resources needs and supplies in each district.
To begin with, we explored the current status of medical resources in all the districts. Figure 3a shows that the number of FCs in the first 4 districts with the severe epidemic in Shenzhen, is the opposite. Especially there are only 5 FCs in Nanshan District with the lowest average number of clinics owned by the ECs, only 0.1, followed by Longhua and Luohu District of 0.17 and 0.19. Then the significant contradiction and imbalance between supply and demand of medical resource allocation are further revealed based on the cumulative opportunities of visiting FCs. Figure 3b indicates that 45 communities with 0 opportunities of visiting FCs are mainly distributed in Nanshan, Longgang, Longhua, Baoan, and Guangming Districts. In mainland China, FCs affiliated with the emergency departments (EDS),the setting it is related to hospital service capacity (standard). Of FCs released by the Shenzhen government, the tertiary hospitals secondary, primary, and below hospitals are accounted for 71.4%, 20.4%, and 8.2% respectively. Moreover, about 62% of FCs in secondary and above hospitals are located in Futian, Longgang, and Baoan Districts. Thanks to that the initial site selection scheme was constrained by the service capacity and location of graded hospitals, it was difficult to achieve equitable access to medical treatment for residents in different communities, thus the location of FCs was not reasonable.
Next, based on the cumulative opportunities of visiting FCs within the ECs in 30 minutes, 3 social groups can be captured (Fig. 4). Among them, the community with 0 opportunities of visiting in 30 minutes is a "low-satisfied community" (22.06%), scattered in every district, the average travel time for them is 42 minutes. In communities with no greater than 3 cumulative opportunities of visiting FCs, "medium-satisfied communities" account for 59.8% with 20 minutes of average travel time. High-satisfied communities are concentrated in Futian, Baoshan, and Luohu District, accounting for 18.14%, with 4 or more cumulative opportunities of visiting FCs and the 15 minutes of average travel time. This illustrates that 40.2% of FCs need to be further optimized to improve medical accessibility and reduce transmission of COVID-19.
Optimized allocation of FCs and effect evaluation
Considering that there are more than 218 hospitals and 661 community health centers in Shenzhen that have not been included as FCs for the coronavirus epidemic, combined with that current FCs need to return to normalize its medical services as the outbreak is stable, it is practical to dynamically adjust the distribution of the fever clinics. The LSCP (Location Set Covering Problem) model38, 43 which attempts to locate a minimum number of servers to cover all demand nodes within the time standard, can achieve the goal of optimizing the rational and balanced use of medical resources and medical equity. Accordingly, we propose two optimization schemes: (i) adding secondary and above hospital as alternatives, (ii) adding primary and above hospital as alternatives. In order to make maximum utilization of medical resources, if the new scheme cannot completely cover the current low-satisfied ECs during the dynamic adjustment process, we will set community hospitals to provide medical services for it. Afterward, we calculated the minimum arrival time-based on the Baidu map API, and finally build the LSCP model to obtain two allocation schemes of FCs.
Table 1 shows that compared with 49 FCs before adjustment, the number of FCs in schemes 1 and 2 have become 32 and 35, respectively. Among them: on the basis of guarantee accessibility, the number of tertiary and above hospitals has decreased by 22 and 18 respectively, which has played an active role in getting medical order to normal.
Table 1
Allocation schemes of FCs and comparison of changes in coverage levels
|
Items
|
Current scheme
|
Scheme 1
|
Scheme 2
|
Hospitals levels with FCs
|
Tertiary
|
35
|
17
|
13
|
Secondary
|
10
|
14
|
10
|
Primary
|
2
|
1
|
12
|
Non-level
|
2
|
0
|
0
|
Community hospitals
|
0
|
28
|
19
|
Low-satisfied community
|
number
|
45
|
28
|
19
|
travel time
|
37
|
42
|
38
|
Medium-satisfied community
|
number
|
122
|
165
|
174
|
travel time
|
19
|
20
|
16
|
high-satisfied community
|
number
|
37
|
11
|
11
|
travel time
|
16
|
15
|
14
|
Specifically, there are 122 medium-satisfied communities in the current scheme, with a coverage rate of only 59.8% and the average travel time is 22.42 minutes (Fig. 4a). After the optimization that retaining 21 original FCs and adding another 11 secondary and above hospitals in scheme 1 (Fig. 4b), the coverage rate of community medium- and high-satisfied hospitals are up to 80.88%, and the average time of going to the FCs is 22.75minutes; In the scheme 2(Fig. 4c), only 15 original FCs are reserved while 20 hospitals are newly added. As a result, the coverage rate of the medium- and high- satisfied community reach 85.29%, and the average time of going to the FCs is 17.94 minutes. To sum up, by adding secondary and above hospitals as alternatives, scheme 1 can improve the fairness of visiting in ECs, transferring the high- and low-satisfied communities to medium-satisfied communities; By choosing hospital at a low level such as primary and above hospitals, scheme 2 can achieve a higher coverage rate, reduce the travel time and improve accessibility.
In addition, for the ECs that can't be covered, 28 and 19 community hospitals are required to provide services, suggesting that primary medical institutions should play a positive role in confronting the virus. Wuhan, the center of the epidemic in China, has adopted similar measures to achieve rapid screening of patients with fever or respiratory symptoms.
Apropos of the COVID-19, the allocation of medical resources must consider the balance and difference between different regions. Inequality in health care comes at the expense of individuals and society, and this will be reflected in inequality in the outcome of the epidemic44. Figure 4a demonstrates that high-satisfied communities are mainly distributed in Bao'an, Futian, and Luohu Districts, which are in the core urban areas of Shenzhen. In contrast, low-satisfied communities are usually located in borders of districts lacking medical resources. In the meantime, the migrant worker after the lunar new year in 2019 in Shenzhen is the highest, accounting for 4.54%, and it will experience large population movements. So that may result in the recurrent spread of the COVID-19 if the medical accessibility of suspected cases is not effectively improved. Through the dynamic adjustment of FCs, the gap between medical accessibility in high-satisfied and low-satisfied communities has been greatly reduced. Supported by the active involvement of primary medical institutions, the ECs that cannot be covered within 30 minutes have been effectively covered, which is necessary for the disadvantaged communities in the junction of the districts. During the rapid spread of the epidemic, many cities have made an immediate reaction and quickly provide a large number of fever clinics (for example 392 FCs in Chengdu and 426 FCs in Chongqing). In addition, these cities are dynamically optimizing and moderately reducing the FCs according to the situation of the epidemic (for example, the number of FCs in Beijing has been adjusted from 89 to 76).