Based on the age profile of the new Covid19 cases reported every week in Sweden; and combined with a couple of assumptions to overcome data limitations, how could these cases translate into deaths in the near future? 👇 
We know that age is probably one of the most important factors when it comes to translating Covid19 case curves into death curves. If only young people get infected we can have plenty of cases but zero deaths.
The Swedish FHM doesn’t report details on each case individually but every day they update the case and death count by age group. With that history and an assumption on case-to-death report delay (4 weeks for example) we can see how the CFR for each age has evolved over time.
These CFR have been very variable in part because the data is by date of report and not actual; and also because of reporting corrections (occasional negative deaths for instance). But we can still take a range of CFR values per segment and calculate a range of outcomes.
Without forgetting the large limitations we can see how despite the recent growth in cases we should only expect a slower increase in weekly deaths in the next 4 weeks, ending at 33-66 deaths on week 43.
Obviously this is a very limited analysis and should be taken as such. Examples of facts that can significantly change the outcome: CFRs deviating from historical ranges (due to changes in testing for instance), changes in case-to-death delays or changes in reporting delays.
Any ideas or suggestions on how to improve this analysis will be more than welcome. Trolls and haters will be welcome as well, this is Twitter after all :)
The idea of using age-stratified CFR values and apply them to latest reported cases comes once again from the work of @zorinaq on Florida epidemic, but then the quality&detail of both the data and analysis in my case are way below that one :)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
