Considering the relationships among cases, we constructed the disease transmission networks and presented the transmission network graphs for the COVID-19 epidemic through a visualization technique based on the individual reports of epidemiological data. Then, in a framework of intuitive and quantitative analysis, we compared the transmission characteristics of COVID-19 of Tianjin and Chengdu in China. This valuable application of the visualization technique was further explored, including tracing the source of infections, discovering potential super-spreaders, and evaluating prevention and control measures. Meanwhile, we discussed the potential insufficiency in the current form of individual epidemiological data. Our research may provide an important basis for jointly constructing multiregional and large-scale disease transmission networks.
Finding “patient zero” plays an important role in preventing the COVID-19 epidemic, such as identifying the origin and further spreading, as well as in studying the transmission characteristics,13-15 as does the identification of the super-spreading event (i.e., one COVID-19 case produces at least 816, 1017, or many more than the average number of secondary patients).18,19 However, epidemiological case reports are written by health workers after investigations with confirmed cases. In each report, only the patient-related source of infections can be obtained from the collected exposure information, which cannot provide enough information to trace back to “patient zero” and present the full transmission chains. Thus, such reports need to be integrated using contact tracing analysis to construct transmission graphs, which provide crucial clues for the identification of “patient zero” and super-spreaders.20,21 For example, by contact tracing and constructing disease transmission chains, researchers found that the epidemic of COVID-19 in Italy had spread much earlier than February 20, 2020, when the first case was confirmed.8 In our study, we found that one COVID-19 patient generally directly produces 0 to 4 infections in Tianjin, while in Chengdu, the number is up to 3. There was no evidence of super-spreaders in the two cities. Terminal nodes of the transmission chains can be applied in several aspects for prevention and control, including identifying potential high-risk populations, determining priorities, and narrowing the scope of quarantine and thus allocating limited health resources effectively.
Currently, the available epidemiological data of each individual case vary from city to city. The exact exposure history can be extracted from released epidemiological data in Tianjin, while in Chengdu, only whether some of the cases have been in close contact with confirmed cases can be extracted, and the relationships among cases are not clear. From January 21 to February 22, 131 (97.04%) patients in Tianjin can be integrated into the transmission network, in which the source, relevant infections and terminal nodes can be revealed. In Chengdu, only 98 (68.53%) cases had infection pathways. The other 45 patients were noninformative cases, and thus, the sources and potential infectious ranges remain unclear, as these noninformative cases cannot be integrated into any transmission chains. By comparing the results of the disease transmission networks of the two cities, we found that the transmission network in Tianjin was more complete with clear transmission chains. Preventive measures can be carried out mainly by focusing on the close contacts of each node, which could reduce the consumption of limited health resources. Noninformative cases and cases with vague information also suggest the risk from unclear transmission chains. In Chengdu, the tracing transmission chains of approximately one-third of the cases were not available. Therefore, the transmission network graph of the COVID-19 epidemic in Chengdu had less coverage and provided less information. The number of nodes in each pathway and the close contacts of each node in the transmission network cannot be determined, indicating higher unpredictable risks in Chengdu. This result indicates that, in epidemiological investigations, the exposure history of each infected should be collected as completely as possible. Traceable transmission chains of each case would greatly reduce the unpredictable risks for prevention and control and avoid the waste of health resources. In addition, the composition of the transmission chains presents the main type of local transmission, which suggests that further prevention and control should focus more on imported or community-spread cases. The length of transmission chains partly suggests that the timeliness of case detection, as well as the quality of the epidemiological investigation reflected by the rate of case coverage, provides a valuable index for evaluation of the efficiency of the local control measures. The relatively poor quality of epidemiological data in Chengdu may suggest the shortage of public health manpower, which may provide evidence for adjusting control and prevention strategies and allocating resources.
Currently, most epidemiological investigation reports with information on exposure behaviors and contacts are provided by unstructured reports with different forms in different cities. Therefore, it is difficult to construct an integrated cross-regional transmission network and gain full use of the epidemiological data for COVID-19 prevention and control. Thus, we suggest that health administrations develop a standard guideline for epidemiological data collection, and all such data should be managed and released in a timely manner.22 On the one hand, researchers can jointly construct a multiregional transmission network to trace the spreading of COVID-19. On the other hand, integrated transmission networks can improve public awareness of COVID-19 epidemics, enhance public compliance with control measures, and reduce the difficulty of implementation and resource consumption. Moreover, with transmission networks, network-based analysis can be carried out to evaluate the transmission rates and the complexity of network structures, which may provide clues for large-scale interventions. In addition, for emerging infectious diseases, constructing transmission chains through contact tracing can estimate infectivity at an early stage to quantify the risk and trends of infectious diseases.23-25
It is worth noting that COVID-19 cases in one category might be included in another category in transmission network graphs. For instance, some cases in the category of family clusters are contained in the category of exposure to infections directly related to Hubei Province. To fully demonstrate the information contained in the epidemiological data, however, this study classified these patients into another category, with family aggregation as a vital feature of infectious disease transmission. Better classification strategies are needed to discuss the local transmission characteristics in different cities. Meanwhile, regional unification should be considered for the classification standard to guarantee the exchangeability of data when cross-regional transmission networks are constructed.
For infectious diseases with stronger infectivity, longer incubation periods and higher fatality risks, such as COVID-19, if public health interventions were not carried out in a timely and effective manner, cases would increase rapidly, consume limited clinical resources quickly and lead to high mortality.26 Therefore, the emphasis of control measures should not only focus on clinical treatments but also ensure sufficient resources in epidemiological investigations.27 These measures could contribute to controlling the source of infections, reducing the risk of exposure, decreasing the incidence by shortening transmission chains, and easing the pressure of clinical treatments. The epidemiological data of the COVID-19 epidemic in Tianjin and Chengdu were used to propose an analysis framework for the individual epidemiological data. Our results illustrated the importance of visualized epidemiological transmission networks in preventing and controlling the epidemic of COVID-19. Currently, the content and format of epidemiological data are not unified, causing the transmission network graphs of Tianjin and Chengdu to show different performances in risk assessment. Therefore, the collection, management, and release of epidemiological data should be improved for the joint construction of large-scale and multiregional disease transmission networks to provide a better understanding of the COVID-19 epidemic and to provide evidence for local prevention and control policymakers.