Read on Twitter
Imagine all the work that goes into collecting and publishing this data, transparently and free of charge, only to have people completely misuse it.
Today's lesson is "cause, effect, and pattern recognition." https://twitter.com/shaynekrige/status/1334105006971170826
Today's lesson is "cause, effect, and pattern recognition." https://twitter.com/shaynekrige/status/1334105006971170826
It's actually really simple: The best chance for a COVID-19 patient to recover, is to receive treatment in hospital, with the right medications and equipment.
The best way to ensure enough beds are available is to prevent the hospitals from filling up.
The best way to ensure enough beds are available is to prevent the hospitals from filling up.
The best way to prevent hospitals from filling up are reducing, as much as possible, the input causes:
* Elective surgeries and non-critical procedures (already done)
* Most things to do with alcohol
* Most things to do with large crowds/spreader events
* Elective surgeries and non-critical procedures (already done)
* Most things to do with alcohol
* Most things to do with large crowds/spreader events
How do you do that? Any range of options from:
* Asking really nicely ("public appeals")
* Making basic hygiene mandatory (masks/sanitizer)
* Putting some restrictions in place
* Putting more restrictions in place
* Locking down problem behaviors
* Asking really nicely ("public appeals")
* Making basic hygiene mandatory (masks/sanitizer)
* Putting some restrictions in place
* Putting more restrictions in place
* Locking down problem behaviors
We've been doing some measure of this for a while now. It's worked. We have free beds in hospitals and people aren't dying.
But we know that there are more cases on the way. The same source above also shows this - a spike in the running total of cases since mid-November:
But we know that there are more cases on the way. The same source above also shows this - a spike in the running total of cases since mid-November:
New cases take around 2 weeks to "end" - either in recovery, or fatality.
To look at the deaths - the output metric in a long, complex process with multiple factors - and conclude that it's not increasing *now* therefore mitigations are unnecessary, is completely dishonest.
To look at the deaths - the output metric in a long, complex process with multiple factors - and conclude that it's not increasing *now* therefore mitigations are unnecessary, is completely dishonest.
If COVID were an instant, same-day killer, that would be one thing. The fact that it can take 2+ weeks to resolve, that you need access to limited amounts of specialized care, and that registered cases are all increasing is every reason to double down on what works.
It's basic cause-and-effect forecasting, and it isn't hard to do. In fact, we've literally done this before - we have real data from real outbreaks showing what the trends look like, and caution should follow a rise in infections. This isn't new.
Or in BI terms:
Deaths are a lagging metric. It tells you what's already happened.
Cases, rate of increase, availability of healthcare, adherence to hygiene requirements and testing are all leading metrics. They indicate what might happen in future.
Deaths are a lagging metric. It tells you what's already happened.
Cases, rate of increase, availability of healthcare, adherence to hygiene requirements and testing are all leading metrics. They indicate what might happen in future.
Again, same source of data as the first tweet. Case trends by sub-district. Most of them show spikes - the Garden Route very clearly, but also Eastern, Tygerberg, Southern.
Cherry-picking a chart that backs up your narrative, from a larger set of data, is profoundly dishonest.
Cherry-picking a chart that backs up your narrative, from a larger set of data, is profoundly dishonest.
