key: cord-1015023-j179quof authors: Yu, Bin; Chen, Xinguang; Rich, Shannan; Mo, Qiqing; Yan, Hong title: Dynamics of the coronavirus disease 2019 (COVID-19) epidemic in Wuhan City, Hubei Province and China: a second derivative analysis of the cumulative daily diagnosed cases during the first 85 days date: 2021-02-06 journal: Glob Health J DOI: 10.1016/j.glohj.2021.02.001 sha: 69344591ac36287e71e95d5e6609c2a17670e151 doc_id: 1015023 cord_uid: j179quof BACKGROUND: Controlling the coronavirus disease 2019 (COVID-19) epidemic requires information beyond new and cumulative cases. This study aims to conduct an in-depth analysis by geographic strata: Wuhan City (hereafter referred to as Wuhan) only, Hubei Province (hereafter referred to as Hubei) excluding Wuhan, and China excluding Hubei. METHODS: Daily cumulative confirmed COVID-19 cases between December 8, 2019 (the date of symptom onset based on patients’ recall), and March 1, 2020, from official sources and published studies were analyzed. The second derivative model was used for information extraction. Data analysis was conducted separately for the three strata. RESULTS: A total of 80 026 diagnosed COVID-19 cases were reported during the first 85 days of the epidemic, with 49 315 cases from Wuhan, 17 788 from Hubei excluding Wuhan, and 12 923 from China excluding Hubei. Analytical results indicate that the COVID-19 epidemic consists of an Acceleration, a Deceleration, and a Stabilization Phase in all three geographic strata, plus a Silent Attack Phase for Wuhan only. Given the reported incubation period of 14 days, effects of the massive anti-epidemic actions were revealed by both the Acceleration and Deceleration Phases. The Acceleration Phase signaled the effect of the intervention to detect the infected; the Deceleration Phase signaled the declines in new infections after the infected were detected, treated and quarantined. CONCLUSION: Findings of the study provide new evidence to better monitor the epidemic, evaluate its response to intervention, and predict the trend long. In addition to re-evaluating the control of the COVID-19 epidemic in China, this study provided a model for monitoring outbreaks of COVID-19 in different countries across the world. The control of the coronavirus disease 2019 (COVID-19) epidemic in China, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2; previously provisionally named 2019 novel coronavirus or 2019-nCoV) has gained much attention across the globe. 1 The first COVID-19 case was reported on December 8, 2019 (the first date of symptom onset based on the patients' recall), in Wuhan City ( hereafter referred to as Wuhan), 2 the Provincial Capital of Hubei Province (hereafter referred to as Hubei) in Central China. As of March 1, 2020, there has been a total of 49 315 (61.6%) confirmed cases in Wuhan only, 3 plus 17 788 (22.2%) cases in Hubei excluding Wuhan, and 12 923 (16.1%) in China excluding Hubei, with a total of 80,026 cases in China. 4 Assessment of control and forecast future of the COVID-19 epidemic need further research to inform decision-makers, medical and public health professionals, and the general public. China has taken a series of massive anti-epidemic actions to control the COVID-19 epidemic (see more details in Appendix Table 1) . Typical examples include but are not limited to the declaration of human-tohuman transmission and outbreak on January 20, the lockdown of Wuhan as a massive epidemic control measure on January 23, delivery of a large number of molecular testing kits on January 26, the Central Government Meeting on COVID-19 led by President Xi Jinping on January 25, the establishment of the National Anti-COVID-19 Leadership Group led by Premier Li Keqiang, and construction of two Emergency Hospitals, Huoshenshan and Leishenshan in days. [5] [6] [7] [8] All these actions may potentially affect the epidemic dynamics of COVID-19 since the epidemic is nonlinear and chaotic in nature. 9 We previously applied a new method, second derivative modeling that successfully characterized the dynamics of COVID-19 transmission in China as a whole in the first two months, evaluated the massive antiepidemic measures, and predicted the trend of the epidemic pattern. 9 In this study, based on our previous study, we aim to investigate the COVID-19 epidemic by three different geographic strata, including Wuhan, Hubei excluding Wuhan, and China excluding Hubei. The cumulative numbers of confirmed cases of COVID-19 on a daily basis were used. The data were derived from two sources: (1) Data from December 8 of 2019 (the first date of symptom onset based on patients' recall) to January 20 of 2020 were derived from a study by Li and colleagues. 2 This was the period before COVID-19 was officially declared as an outbreak. (2) Data from January 21 to March 1, 2020, were derived from the daily reported and officially finalized cases. Data for Wuhan City and Hubei Province were derived from the Wuhan Municipal Health Commission 8 and data for China were derived from the National Health Commission of P. R. China. 10 To address the purposes of this study, we derived daily data: (1) for Wuhan only, (2) for Hubei excluding Wuhan by subtracting the daily cases of Wuhan from the provincial data, and (3) for China excluding Hubei by subtracting the cases in Hubei (including Wuhan) from the national data. Detailed data were included in Appendix Table 2 . To characterize the dynamics of the COVID-19 epidemic, identify evidence of massive anti-COVID-19 actions on dynamic changes in the epidemic, and predict future trends at the earliest time possible, we took the second derivative modeling approach. 9 Using this modeling approach, we first described the daily cumulative number of cases ( ) as: where represents new cases at day i (i=1, 2, 3, … t). Taking the first derivative of ( ), we obtain ( ): where ( ) is equivalent to the daily new cases. It provides information regarding the speed of the epidemic on a daily basis. Although new cases are updated daily during the epidemic, this measure is sensitive to changes in disease diagnosis and is not informative to signal future trends. Therefore using ( ) alone is not adequate to monitor and make inferences about changes in epidemic dynamics, and inform decision-makers, medical and health professionals, as well as the general public about the epidemic. To increase sensitivity to detect changes in the epidemic and to reduce potential impact from changes in diagnostic criteria, we took the derivative of ( ) to obtain the second derivative ( ): where ( ) measures the speed of changes in daily new cases or acceleration of ( ). The first COVID-19 case was reported in Wuhan on December Instead of reported daily cases, ( ), the first derivative of ( ) for the three geographic strata is presented in Fig. 1 . The ( ) for Wuhan (red line) indicates a long-term slow growth of the epidemic after the first case on December 8, 2019, until January 27 when a rapid increase forming a spike on the day after the massive number of testing kits was distributed to Wuhan on January 26. The ( ) reached a peak with a total of 13,436 new cases reported on February 12, the day when the diagnostic criteria were revised to include all patients with positive clinical symptoms regardless of laboratory confirmation. After the peak, the ( ) declined rapidly to below 400 on February 25 and left off at around 400 thereafter before declining again on Data were derived from one published paper by Li and colleagues, 2 Wuhan Municipal Health Commission, 8 Commission of the P. R. China. 10 December 8, 2019 is the date of symptom onset based on patients' recall. In this study, we presented the results from the second derivative modeling analysis of the cumulative cases of COVID-19 by geographic strata in China. It is difficult if not impossible to determine ahead of time when, where, and whom an outbreak of an epidemic like COVID-19 will attack. The best approach is to closely monitor and capture information at the earliest time possible to inform decision-makers, medical and health professionals, and the general public. Findings of the study provide new evidence without extra data but essential for us to better understand the outbreak and the current status of the COVID-19 epidemic as well as its future trend in a real-time manner. Findings of the study also add data supporting one of our previous studies that second derivative modeling can be used as a surveillance tool to monitor the outbreak of infectious disease epidemic and inform decision-makers for massive anti-epidemic actions, and evaluate the response of the epidemic to intervention and forecast the future based on current surveillance data not only in China but different countries across the world. 9 In this study, we demonstrated that the second derivative model, built using data on daily cumulative cases that are widely available, can extract the hidden information to better inform us about the dynamics of COVID-19. Taking the epidemic in Wuhan as an example, information derived from cumulative daily cases can better inform us about the epidemic. For instance, the first spike on January 27 was attributed to the distribution of a large number of testing kits on January 26, and the spikes on February 12 and 13 were attributed to the change of diagnostic criteria which were also observed in the first derivative results. Further, the results of the second derivative model indicated a unique four-stage epidemic of COVID-19 in Wuhan which can be observed neither in cumulative cases nor in new cases. In addition to the Silent Attack Phase before the declaration of the outbreak, the 15-day Acceleration Phase from January 21 to February 4 was reflective of an increase in the epidemic from the day when the massive interventions were activated on January 21 to February 4, one incubation period after the intervention. 9 The 14-day Deceleration Phase from February 5 to 19 was reflective of another incubation period from February 5 when the massive anti-epidemic actions actually showed their effects. After the Deceleration Phase, the epidemic entered the Stabilization Phase when most infected cases were detected and treated/quarantined. This phase indicates that the epidemic was at a very low level, suggesting the success in controlling the COVID-19 epidemic in Wuhan. Similar to the situation in Wuhan, results of second derivative modeling for Hubei excluding Wuhan and China excluding Hubei also revealed responses of the COVID-19 epidemic to the massive anti-epidemic actions, as indicated by the spike in the second derivative on February 12 in Wuhan and Hubei, corresponding to the day when a large number of diagnosed cases with clinical symptoms but not confirmed with a blood test, 11 and the spikes on February 20 and 21 in China excluding Hubei, corresponding to the inclusion of diagnosed cases in prisons. [12] [13] In addition to these spikes, the second derivative results suggest three similar phases as Wuhan, including an Acceleration Phase, a Deceleration Phase, and a Stabilization Phase; but the duration of the three phases differed a bit across the three geographic strata. The two strata outside of Wuhan entered the Deceleration Phase and Stabilization Phase a few days earlier than Wuhan. This could be due to the effect of the massive anti-epidemic action of city lockdown. Despite much social, economic, and political influences, evidence from our analysis suggests that massive anti-epidemic action, like a citywide lockdown, would be the only option at the time when an epidemic has already reached its later stages. To avoid the challenging city lockdown strategy for future interventions, potential actions should be taken in the early stage, including, but not limited to, informing the public about social distancing, hygiene, and other common approaches of prevention in an earlier stage although the CDC and health professionals may not have adequate evidence of the disease at that moment, applying the data science and machine learning technique to actively collect social media information about outbreaks, infections and symptoms as a supplement to the official report from the hospitals and local CDCs, and building professional teams (including psychologists, psychiatrists, social workers, etc.) to take care of people's mental health issues (e.g., anxiety, depression, etc.) as well as other society and community-related problems. As in a previous study for China as a whole, 9 results from the second derivative of cumulative daily cases in this study can better inform us about future trends of the COVID-19 epidemic in Wuhan, Hubei excluding Wuhan and China excluding Hubei. For example, the second derivative in Wuhan entered the Stabilization Phase on February 20, with one incubation period, it is expected that on March 6 the epidemic may substantially decline. It is worth noting that the ability to predict future trends for a subpopulation is slightly weaker than the full population. For example, in our previous study with data for the whole of China, it was predicted that the epidemic would be under control by the end of February. 9 However, the method with 14-15 days of incubation period worked only for Wuhan, but not for Hubei excluding Wuhan and China excluding Hubei. One explanation for this could be due to the cross-boundary movement of the infected persons. The control and prevention of the pandemic of COVID-19 require more knowledge about the epidemic and the effects of taken actions. Findings of the study indicated that the citywide lockdown of Wuhan successfully flattened the curve outside Wuhan, and the epidemic curve in the regions of China excluding Hubei Province and Hubei excluding Wuhan City entered the Stabilization Phase earlier than Wuhan. Findings of the study provide new evidence to better monitor the epidemic, evaluate its response to intervention, and predict the trend long. In addition to re-evaluating the control of the COVID-19 epidemic in China, this study provided a model for monitoring outbreaks of COVID-19 in different countries in the era of the COVID-19 pandemic. This study depends solely on reported data. Although the second derivative of an epidemic is not very sensitive to changes in diagnostic criteria, the method will work better without such change. More work is needed to inform decision-makers, medical and health professionals, and the general public to understand and use the method in controlling the COVID-19 and other similar epidemics in the future. Despite these limitations, the study provided important data and informative evidence to characterize the epidemic of COVID-19 by geographic strata in China. Findings of the study regarding the Acceleration, Deceleration, and Stabilization phases provide Coronavirus disease (COVID-19) pandemic. World Health Organization website Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia Wuhan Municipal Health Commission. 2019-nCoV situation report National Health Commission of the People's Republic of China Novel Coronavirus (2019-nCoV) technical guidance: Laboratory testing for 2019-nCoV in humans Chinese Center for Disease Control and Prevention. Epidemic update and risk assessment of 2019 Novel Coronavirus Wuhan Municipal Health Commission. Updates of the prevention and control of 2019-nCoV First two months of the 2019 Coronavirus Disease (COVID-19) epidemic in China: realtime surveillance and evaluation with a second derivative model National Health Commission of the People's Republic of China. Updates of the COVID-19 epidemic Hubei Province novel coronavirus pneumonia updates on Zhejiang Province novel coronavirus pneumonia updates The State Council Information Office of the People's Republic of China. Shangdong Province novel coronavirus updates and control press release Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia Wuhan Municipal Health Commission. Updates of the prevention and control of 2019-nCoV ) National Health Commission of the People's Republic of China. Updates of the COVID-19 epidemic *The date of symptom onset based on patients' recall. # The number of cumulative cases of Hubei excluding Wuhan was calculated as the cumulative cases in Hubei Province subtract the cases in Wuhan City. In theory, the number of cumulative cases should increase along with the date. However, there may exist some issues in the official reports The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.