Wuhan is the capital city of Hubei Province and a major transportation hub in central China with a population of around 11.2 million (about 14% of them are aged 65 or over). By the end of 2018, Wuhan had 6,340 health institutions (including 398 hospitals), 10.8 beds per 1,000 citizens, and 4.5 physicians per 1,000 citizens (14). However, like the overall situation in China, the health care delivery system of Wuhan is fragmented, hospital-centred, and treatment-dominated, with little effective collaboration among institutions in different tiers(detail in supplementary (Box 1) (15, 16). Additionally, the lack of a mandatory primary care system via general practitioners and quality medical resources mainly occupied by hospitals allowed residents of Wuhan to have unrestrained access to second or tertiary health services. This further exacerbated fragmentation and lowers the efficiency of the health care delivery system, particularly where minor health concerns are experienced. In such cases, where primary care services would perhaps suffice, specialized health care is being unduly overwhelmed and exploited.

We performed a retrospective qualitative study based on the document analysis of COVID-19-related materials and interviews with first-line people in Wuhan. The materials included official policies, technical guidelines, and reports related to COVID-19. The materials were retrieved using terms related to COVID-19 (Chinese words included “新冠”, “新型冠状”, and “肺炎”) from the official webpage at the national level, city level, and community level, published during December 01, 2019 and March 20, 2020. Data were extracted by two investigators (SC and PZ) using a standardized form including the following domains: publication date, objectives/challenges, participants-involved, and the functions and responsibilities of the participants. All retrieved materials were screened independently by two investigators (SC and PZ). We only included materials related to COVID-19. Discrepancies in the inclusion or exclusion of materials during screening were discussed with a third reviewer (YH) until consensus was achieved.

Authors (PZ, YH, and ZL) have been in the front-line of the response since early 2020. Their affiliation is the designated hospital for COVID-19; therefore their first-hand experience was also used in the study. Specifically, with regard to information on designated hospitals, including patient screening, transfer, and treatment procedures, we interviewed hospital leaders and other front-line staff within departments in charge of epidemic prevention and control, including the department of infectious diseases, emergency department, fever clinic, and department of hospital external liaison. For information about Fangcang hospitals, besides the information from author (ZL), who was the deputy director of one of the Fangcang hospitals, we also interviewed the leaders of the national rescue team in Hubei and the leaders of the districts, who were involved in the operation of the Fangcang hospital and the patient transfer process. Regarding the community and isolation points, we interviewed the staff from affiliated community centers and isolation points; information about the deployment of the command center of prevention and control is mainly based on interviews with staff who were second to the command center, including resource allocation, department coordination, and information release at various stages.

The measure of effectiveness involves at least two concepts, namely cost and outcome. There is an article that estimates the corresponding costs (17), but because Wuhan is the first city to face the COVID-19 outbreak and many measures are generated and improved in the process of exploration, so the corresponding costs are much higher than those of truly effective measures. Also, because China uses the power of the whole country to control the spread of the outbreak in the shortest possible time at all costs, effectiveness based on cost and outcome does not have a particularly significant reference value. Here, we just used the outcome to reflect the effectiveness. Similar handling can be found in a publication (2). The outcome was measured by confirmed cases, death cases, and recovery cases of COVID-19 in Wuhan, and data was collected from the official webpage (18).

Initial extracted data were reviewed by all authors, and two themes and four sub-themes were formulated. The two themes are the evolution of Wuhan's prevention and control strategies on COVID-19 and corresponding effectiveness. The four sub-themes are routine prevention and control period, exploration period of targeted prevention and control strategies, mature period of prevention and control strategies, and recovery period. Data were re-coded into a matrix (Table 1 is its compact edition).

All coding was done in Excel. How Wuhan combatted COVID-19 through multi-tiered and multi-sectoral collaboration, overcoming its disadvantaged fragmented, hospital-centered, and treatment-dominated healthcare system, was illustrated and summarized.

Demonstrating effectiveness for each response using a systems approach is challenging in the case of Wuhan as changes in one part (for example, primary health centers) altered other parts of the system (like hospitals) in ways that are difficult to measure. In addition, measures in Wuhan were implemented over short intervals and led to high overlap, making the separation of effectiveness for each response even more complex. Therefore, the judgment of effectiveness was mainly based on the subjective account of front-line personnel: whether relevant measures are important to control the spread of the virus.

There were three milestone for Wuhan's COVID-19 crisis: (i) the date of determining that COVID-19 can be transmitted from person to person (January 22, 2020); (ii) putting the Fangcang hospital into use and providing technical support for community and primary care (February 05, 2020); and (iii) the day that the daily number of new confirmed COVID-19 cases fell below ten (March 11, 2020). Accordingly, we divided the evolution of Wuhan's prevention and control strategies on COVID-19 into four periods. Table 1 summarizes the main participants in each period and their corresponding prevention and control strategies. Figure 1 shows the COVID-19 cases (including confirmed, death, and recovered) in Wuhan in each period.

Period I, routine prevention and control period (Figure 1, from December 27, 2019, to January 21, 2020, 26 days), featured by the formidable COVID outbreak in lack of specific evidence on diagnosis, treatment, prevention, and control strategy.

During this period, COVID-19 was seen as an atypical pneumonia with an unknown pathogen. Workers of disease prevention and control were mainly public health administrative departments and hospitals. After the pathogen was identified as a novel coronavirus, the prevention and control strategies in this period were made based on conventional experience (not targeted) and with the hospital as the primary first-line units, following lack of specialized knowledge.

Period II, exploration period of targeted prevention and control strategies (Figure 1, from January 22, 2020 to February 04, 2020, 14 days), featured by a combination of chaos and exploration.

From January 22, 2020, following the official announcement that COVID-19 is a human-to-human infection disease, Wuhan entered the road of exploring targeted prevention and control strategies. China has experienced many acute infectious disease emergencies, including Severe Acute Respiratory Syndrome (SARS) in 2003 (19), the H1N1 flu epidemic in 2009 (20), the H7N9 avian flu epidemic in 2013 (21), and Middle East Respiratory Syndrome (MERS) in 2015 (22). Drawing from these experiences, China has been gradually improving its health-emergency-related surveillance, preparedness, and response capacities. At the initial point of this period, a top-down temporary directive team was organized, led by the provincial governor and composed of the provincial-city-community level of heads of health, public security, civil affairs, finance, human resources, transportation, economy, information, and news.

The series of anti-epidemic measures implemented during this period were derived from past anti-epidemic experiences, such as social distancing and the establishment of hospitals. This perhaps explains why Wuhan could respond quickly and decisively from when COVID-19 was identified as a human-to-human infection disease. Not all empirical measures applied to this new type of virus, but the determination to control the epidemic at all costs determined the rapid response of Wuhan's epidemic prevention and control, and laid the foundation to identify effective protective measures e.g., social distancing and movement control order, later recommended by the WHO. To further enhance the ability of rapid learning and adaptation, feedback from first-line anti-epidemic measures and related issues were collected and aggregated to the central control team. For instance, to fully implement body temperature screening in various places and strengthen the monitoring of fever clinics in medical institutions, requisite 2-hour direct network reporting, 12-hour feedback of test results, and complete on-site epidemiological investigation within 24 hours were demanded.

The most well-known measure taken by Wuhan is the city-wide lockdown from January 23, 2020. The lockdown of Wuhan confined the outflow of possible contaminated populations and constrained the spread of COVID-19. But the sudden closure also brought a great negative impact on the normal function of the whole city and instituted panic among the citizens. There were three reasons for the serious overcrowding and cross-infections within the hospitals: the lack of sufficient buffer time meant the prevention and control strategies within Wuhan had not been completely synchronized with the closure; Wuhan itself had a fragmented, hospital-centred, and treatment-dominated health care delivery system; and (c) a large volume of individuals (both COVID-19 patients and non-COVID-19 patients) -presented at hospitals with to an irrational fear of virus infection (COVID-19 or seasonal flu). As a result, initially, the city-wide lockdown not only failed to well-control the spread of the COVID-19 within Wuhan, but instead impeded COVID-19 patients getting timely screening and health services. The transient rise in the number of cases between Jan 26 and Jan 29 (Figure 1, phase II) could be partially explained by this failure. Furthermore, the health services for patients with non-respiratory diseases, like women in labour, were also influenced (23).

The negative results above forced Wuhan to adjust its hospital-centred strategies. To meet the needs beyond the COVID-19 patients and to reduce the nosocomial infection, Wuhan re-designed the roles of hospitals in the system response (e.g., some hospitals were designated as infectious disease hospitals admitting COVID-19 patients, while some were designated for treating patients with other urgent or non-urgent conditions), and included community and primary care as the first line to prevent and control COVID-19. In addition, the construction of independent quarantine centres for the infected cases, carried out almost at the same time as the city-wide lockdown, was gradually completed. Up to February 01, 2020, a multi-tiered and multi-agent prevention and control strategy has begun to take shape. However, cooperation between agents was still absent at this time, because the functional role of each agent had not yet been confirmed. For instance, community and primary care at this time mainly helped to measure body temperatures, and were far from fulfilling their first-line role.

According to the available evidence at this stage, on February 02, 2020, Wuhan clarified four special groups of the population who needed to be screened and monitored for COVID-19, namely fever patients, their close contacts, suspected COVID-19 patients, and confirmed COVID-19 patients. Meanwhile, the functional role of the community and primary care and independent quarantine center was initially determined. Community and primary care were positioned to help classify and manage the above four types of population. Independent quarantine centers were divided into two categories (“admission” and “to be discharged”) to reduce the cross-infection within and the pressure on the hospital. The “admission” quarantine center only accepted patients with suspected fever patients or confirmed positive patients, and the “to be discharged” quarantine center only accepted patients who were discharged from the hospital after a negative nucleic acid test. By February 02, 2020, a multi-tiered and multi-agent cooperation prevention and control strategy had begun to take shape.

To screen the four key groups of people for COVID-19, Wuhan began to screen the general population for the first time from February 03, 2020. This time, screening encompassed the grand mobilization of the whole city, but still failed because screening (prevention) and control were not fully coordinated. Due to the lack of technical support, the community and primary care still passively fought COVID-19: the main energy of the community and primary care had been spent on checking the numbers of suspected and confirmed patients, and failed to take control of community residents in points of entry, resulting in the movement of people still being common. This failure was also reflected in the comparison between the newly confirmed cases during this time screen (after February 03, 2020) and those in phase 3 (after February 08, 2020, illustrated below) (Figure 1), which was seminal to the COVID-19 defence success in Wuhan. Nevertheless, although a kind of negative result, it laid the foundation for the evolution of the multi-agent cooperation model.

In summary, this period was a combination of chaos and exploration, and Wuhan showed a quick response capacity. During this period, four types of the population were identified, the main body of COVID-19 prevention and control gradually transitioned from simplification to diversification, and the functional role of each agent was initially clarified.

Period III, mature period of prevention and control strategies (Figure 1, from February 05, 2020 to March 10, 2020, 35 days), featured by a well operated multi-tiered and multi-agent cooperation.

From February 05, 2020, Fangcang shelter hospitals had begun to be put into service, and on February 06, 2020, infection control specialists were sent to community and primary care clinics to provide technical support, marking that Wuhan's COVID-19 prevention and control model had entered a mature period. Technical support encompassed vetting environmental characteristics of the community and the composition of residents, developing targeted prevention and control measures, including how to screen, how to educate and raise awareness, how to transfer confirmed patients, how to deal with domestic waste, how to disinfect the community, and how to provide psychological assistance. Figure 2 visualizes the final multi-tiered and multi-agent cooperation model in Wuhan.

In this final model, different from the regular health care delivery system in Wuhan, all referrals of patients now had to go through primary care physicians, and the advanced hospital had the right to refuse patients' over-specialized/advanced needs if they bypassed primary care. Community and primary care services helped to manage and screen the population and classify the four types of the population via physical examinations, routine blood screening, and chest imaging, as well as the provision of health education on COVID-19 prevention. Independent quarantine centers functioned as a buffer between screening and treatment. These centers not only accepted suspected cases from the community and conducted confirmatory nucleic acid tests to determine the final flow of patients, but also accepted recovered cases from the hospital for the second nucleic acid test to ensure a non-infectious state, before the patient returned to the community. Fangcang shelter hospitals provided treatment services to mild confirmed COVID-19 patients and also functioned as a buffer between independent quarantine centers and tertiary hospitals to alleviate the burden on tertiary hospitals, by accepting the mild COVID-19 patients from the quarantine center and the COVID-19 patients from tertiary hospitals whose symptoms had turned from moderate or severe to mild. Public or designated hospitals usually are tertiary hospitals and specifically aim to provide services for moderate or severe COVID-19 patients. Health administration, and Centers for Disease Control (CDC) and Prevention worked as the central command center, ensuring the safe and efficient operation of the entire system, and promptly responded to emergencies by deploying material and human resources.

The cooperation of Wuhan's model in practice combines the integration of both horizontal and vertical cooperation. Figure 2 is a direct demonstration of how different agents in Wuhan collaborate vertically, while horizontal cooperation is mainly reflected by cooperation between each agent and command center, as well as the work within the community and primary care. Through the big data platform, the command center contributed to a whole-of-health and whole-of-society system by integrating the data of hospitals, CDC, community and primary care, medical insurance, polices, civil affairs, communications, and other related departments, which not only facilitates the screening of the population, but also ensures that no contaminated cases are omitted. In addition, the announcement of free COVID-19-related services thoroughly disconnected the long-term criticized revenue-linkage between doctors and patients, and increased people's confidence and compliance with the official anti-COVID-19 decisions (e.g. mask-wearing and maintaining social-distancing).

The horizontal cooperation within each community was mainly done by a team with at least five members, consisting of community medical staff, community officials, community volunteers, and community police. This team together were in charge of about 300 families (800 populations) and responsible for the screening, health education, mental health support, daily living support for special populations such as the elderly living alone and disabled people, and the control of unnecessary mobility of residents. It is worth emphasizing that it is the work at the level of community and primary care that greatly promoted the practical implementation of prevention and control measures. Unlike the initial response to COVID-19, at this stage, people had a certain understanding of this disease (through official media and internal education in the community) and gradually realized the important role of individual-level cooperation in the overall anti-COVID-19 progress. Under the encouragement of the community team, people engaged more in online activities to reduce physical contact with each other.

From February 08, 2020, Wuhan city began to screen the entire population for the second time to distinguish the four key groups of people, and more than 10 thousand people were screened (Figure 1). From February 16, 2020, to February 19, 2020, Wuhan conducted the third all-population screening to ensure that no contaminated population were unidentified. In hindsight, it was the second screening that predicted the arrival of Wuhan's success in fighting the COVID-19, with the third screening consolidated Wuhan's anti-COVID-19 results.

Period IV, recovery period (Figure 1, from March 11, 2020, to present), featured by community and primary care-led COVID-19 campaign.

From March 11, 2020, the daily number of new confirmed COVID-19 cases had fallen below ten (Figure 1), marking that Wuhan's COVID-19 prevention and control strategy had entered a recovery period. During this period, social order and medical order were gradually returning to normal, for instance, independent quarantine centers and Fangcang shelter hospitals were shutting down. However, the prevention and control at the community level had not been loosened but instead been adjusted according to actual needs. The communities were divided into low, medium, and high-risk areas, based on the number of newly confirmed cases. Only a community that satisfied two conditions could be recognized as a COVID-19-free community: a) there were no new confirmed case of COVID-19 for more than 14 days and b) no new case of three types of personnel (suspected COVID-19, fever cases, and close contacts) within 14 days. Communities assessed at different levels of risks adopted different corresponding prevention and control strategies to prevent the rebound of COVID-19 (Table 1).

In addition, for various institutions and places, Wuhan had formulated corresponding strict prevention and control strategies. For instance, operation of entertainment and leisure venues were suspended until June 12 until the daily number of new confirmed COVID-19 cases had fallen below ten from March 11, 2020; in public places such as shopping malls, supermarkets, bookstores, and bazaars, temperature checks were required for entrance, and customers were urged to wear masks; hotels, restaurants and other catering establishments were required to extend table spacing in addition to controlling the number of diners. Furthermore, another two large-scale screenings were carried out to ensure that COVID-19 would not rebound. On April 14th, the Chinese CDC launched the COVID-19 seroepidemiological sample survey in Wuhan, and the sampled number reached 11 thousand. Later, on May 2nd, Wuhan required all communities to complete nucleic acid screening for all members of their jurisdiction within 10 days.

From the date when COVID-19 was confirmed as a human-to-human infection, to the date when the daily number of new confirmed COVID-19 cases reached more than ten thousand, and then to the date when the daily number of new confirmed COVID-19 cases had fallen below ten, Wuhan has experienced 49 days (phase II and phase III, Figure 1), including 14 days of exploration on the specific anti-COVID-19 strategies (phase II, Figure 1). These short periods demonstrate the rapid response of Wuhan to the epidemic and the efficacy of Wuhan's coordinated approach in fighting against COVID-19.

The only obvious pulse in panel 2 of Figure 1 predicted the arrival of Wuhan's success in fighting against the COVID-19. It indicates that Wuhan's coordinated model is not only strong in identifying suspicious COVID-19 cases, but also strong in accommodating and controlling the spread of virus borne by the patients. Up to 15 July 2020, the COVID-19 related mortality in Wuhan is 5.43% (11), lower than the average worldwide (24). Wuhan's low mortality indicates that the coordinated model matches the health needs of COVID-19 patients very well. Finally, Wuhan's coordinated approach has a property of thoroughness in fighting against COVID-19 as there is no rebound after COVID-19 was controlled (Figure 1).