Covid: lethality over time

• Data source: http://www.dkriesel.com/corona, referring to https://data.humdata.org/dataset/novel-coronavirus-2019-ncov-cases (retrieved 22/02/20)
• Data visualized for Germany
• Marked virus variant start dates are taken from https://www.rki.de/DE/Content/InfAZ/N/Neuartiges_Coronavirus/Virusvariante.html (retrieved 22/02/20).
  These dates might not necessarily represent the date, where this variant "took off".
• Marked date for vaccination start is taken from https://github.com/robert-koch-institut/COVID-19-Impfungen_in_Deutschland (retrieved 22/02/20).
• The data is smoothed with a rolling average of 7 days.
• According to https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0257978, figure 1, the mean time between diagnosis and death is about 20 days.
  Lethality is accordingly the ratio between deaths (time-shifted by -20 days) and cases.
  This is obviously a very strong simplification.
• Lethality is plotted with a thick line for periods, where the number of daily deaths was equal or larger than 50, otherwise with a thin line.
  I assume that the lethality value is not very representative for low numbers of deaths, potentially because the estimate for the time between diagnosis and death is then not correct.
• Effects like unreported cases e.g. due to exceeded lab capacities are not considered.