Now-casting the COVID-19 epidemic: The use case of Japan, March 2020

Stephan Gloeckner; Gerard Krause; Michael Hoehle

The comparison of the nCoV-2019 Data Working Group line-list data and the data gathered from the WHO Situation reports for date of confirmation shows similar trends, while the WHO data seems to lag one day behind, starting February 14, 2020. WHO reported for February 5 th a cumulative total of 33 cases, but set the case count back on February 6 th to a cumulate total of 25 cases. We set the cases of February 6 th to unknown and restarted the plotting of the epidemic curve at this day (see Figure 1 ). Our modelling of symptom onsets show a decrease of reporting delay over the time (see Figure 2) , from a median of ten days in calendar week four to six days in calendar week eight. Figure 3 shows the results of the now-cast for the line-list data on March 2 nd 2020 and done for the previous 14 days. Case counts more than nine days ago receive little adjustment, because the delay distribution is such that almost all cases are available. As an example, the number of now-casted cases on February 29 th 2020 is 6.2 on top of the observed 1.9 with a 95% prediction interval of 0.5 to 23.2. From the adjusted cases between February 28 th and March 1 st it becomes clear that despite the observed . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.