Statistical and network analysis of 1212 COVID-19 patients in Henan, China

Abstract

Background: COVID-19 is spreading quickly all over the world. Publicly released data for 1212 COVID-19 patients in Henan of China were analyzed in this paper.

Methods: Various statistical and network analysis methods were employed.

Results: We found that COVID-19 patients show gender (55% vs 45%) and age (81% aged between 21 and 60) preferences; possible causes were explored. The estimated average, mode and median incubation periods are 7.4, 4 and 7 days. Incubation periods of 92% of patients were no more than 14 days. The epidemic in Henan has undergone three stages and has shown high correlations with the numbers of patients recently returned from Wuhan. Network analysis revealed that 208 cases were clustering infected, and various People's Hospitals are the main force in treating COVID-19.

Conclusions: The incubation period was statistically estimated, and the proposed state transition diagram can explore the epidemic stages of emerging infectious disease. We suggest that although the quarantine measures are gradually working, strong measures still might be needed for a period of time, since ∼7.45% of patients may have very long incubation periods. Migrant workers or college students are at high risk. State transition diagrams can help us to recognize the time-phased nature of the epidemic. Our investigations have implications for the prevention and control of COVID-19 in other regions of the world.

Keywords: Aggregate outbreak; COVID-19; Henan province; Incubation period; Network analysis; Time-phased nature of epidemic.

Fig. 1

Epidemic situation in Henan through February 14, 2020. A. Epidemic situation on February 14, 2020; B. The evolution of provincial daily increase in confirmed cases and provincial cumulatively confirmed cases. C. The heatmaps of daily increase in confirmed cases and cumulatively confirmed cases in 18 regions. Here, ZZ: Zhengzhou; KF: Kaifeng; LY: Luoyang; PDS: Pingdingshan; AY: Anyang; HB: Hebi; XX: Xinxiang; JZ: Jiaozuo; PY: Puyang; XC: Xuchang; LH: Luohe; SMX: Sanmenxia; NY: Nanyang; SQ: Shangqiu; XY: Xinyang; ZK: Zhoukou; ZMD: Zhumadian; JY: Jiyuan. Similarly hereinafter.

Fig. 2

Summary statistics on the gender and age distributions of confirmed cases in Henan. A. MFR in the whole province; B. MFRs in the 18 regions. C. Scatter plot of MFR versus MFPR in the 18 regions; D. Age distribution for confirmed cases in the whole province and its 18 regions.

Fig. 3

Statistical results of estimated incubation periods for 483 confirmed patients in Henan. A. Frequency distribution of estimated incubation periods and the probability density curve for fitted incubation period. B. The cumulative frequency distribution of estimated incubation periods. C. Transfer diagram from exposure to infection for the 483 patients. Nodes at the two ends of an edge correspond to the exposed date and the date with clinical symptoms or diagnosis. Edges represent the transfer from the two dates, and thicknesses of edges are proportional to the numbers of patients. Self-loops mean that the dates for exposure and appearance of clinical symptoms or diagnosis were the same. D. Weighted indegree and outdegree distributions of the directed graph as shown in C. Weighted indegree denotes the total number of patients with clinical symptoms or diagnosis; while weighted outdegree represents the total number of exposed persons that will be confirmed to be infected later.

Fig. 4

COVID-19 infection and Wuhan travel histories. A. Bar-plots for the numbers of four categories of patients in the 18 regions of Henan. B. Transportation manners of 250 patients that came from Wuhan. C. The heat-map of the correlation matrix among Y, $X_i(i=\mathrm{1,2,3,4})$ (corresponding to 1-5). D. The scatter plots of Y versus X₁-X₄ respectively and the corresponding fitted linear regression lines.

Fig. 5

Network analysis among patients, hospitals and their relationships. A. A heterogeneous network that contains 1105 patients, 248 hospitals, 123 inter-hospital transfer relationships that involved 206 patients and 208 clustering infected patients. Nodes include patients and hospitals. Edges among patients represent relationship between relatives, friends or colleagues; Edges between patients and hospitals denote the patients were treated in the related hospitals; while edges among hospitals indicate the inter-hospital transfer treatments. Sizes of nodes were proportional to nodes’ degree. The thickness of edges was proportional to the numbers of patients. B. The heterogeneous network for patients in Kaifeng. C. The top-20 hospitals with the most patients in treatments. D. Distributions of the 208 aggregate outbreak patients in the 18 regions.