5 ms·

You write an article mixing all kinds of risks together that require careful individual analysis, discuss different kinds of outcomes (hospitalization, death, l

You write an article mixing all kinds of risks together that require careful individual analysis, discuss different kinds of outcomes (hospitalization, death, long covid), discuss different kinds of groups where my first reaction would be to see if the cutoffs are carefully chosen to support the headline - this is a fucking mess. And then you end it all with “The booster policy is the booster policy". My god, science reporting in the US is dead.

wolf550e 5y ago
Why don't they write/draw a Sankey diagram with something like 1 million people on December 1st 2019, with age bracket percentages, and then draw how some got COVID, some got hospitalized, some died, some got the vaccine, some got a booster, some died from things that are not COVID, all in proportion to how it happened in US population. The width of the flows would make it clear how effective vaccines are.
- yuliyp 5y ago
  Trying to draw a Sankey diagram that accurately describes what is happening are quite difficult: many of the numbers for those < 65 years old, and especially for those < 50 years old) are so small as to be hard to read: telling 1/1k vs 1/100k in a Sankey diagram doesn't work. They're both 1 pixel wide.
- tomrod 5y ago
  Sankey doesn't work well because we moved from SIR to SEIR+S with Delta -- meaning that we had breakthrough infections and the potential for waning immunity. Omnicron continues this. [0] https://en.wikipedia.org/wiki/Compartmental_models_in_epidemiology#The_SEIS_model https://en.wikipedia.org/wiki/Compartmental_models_in_epidem...
alea_iacta_est 5y ago
2 years in, we don't even have the numbers of people hospitalized from covid or with covid. It's not science reportings that is dead, it's science itself.
- nradov 5y ago
  Currently about half of COVID-19 patients in hospitals are incidental cases who were admitted for another reason and then tested positive during admission screening. https://gothamist.com/news/new-preliminary-state-data-shows-half-people-nyc-hospitals-covid-19-were-admitted-other-reasons https://gothamist.com/news/new-preliminary-state-data-shows-...
- tomrod 5y ago
  There are many people who have these numbers and study them. Public health officials in the US have both the reported numbers as well as piggy backing on the system set up for influenza monitoring (CLI & ILI). Simply because not all public reporting is as rich as, say, Illinois or Pennsylvania's, doesn't mean these figures aren't carefully scrutinized -- especially when threatened by political leaning from either side of aisle (such as when the then-current administration pressured the CDC to change and alter reports). In my experience, hospital operations officers, health information exchanges, and, depending on the state, state HHS agencies don't mess around. [0] https://www.cnbc.com/2021/04/09/trump-officials-bragged-about-pressuring-cdc-to-alter-covid-reports-emails-reveal-.html https://www.cnbc.com/2021/04/09/trump-officials-bragged-abou...

wolf550e 5y ago

Why don't they write/draw a Sankey diagram with something like 1 million people on December 1st 2019, with age bracket percentages, and then draw how some got COVID, some got hospitalized, some died, some got the vaccine, some got a booster, some died from things that are not COVID, all in proportion to how it happened in US population. The width of the flows would make it clear how effective vaccines are.

yuliyp 5y ago
Trying to draw a Sankey diagram that accurately describes what is happening are quite difficult: many of the numbers for those < 65 years old, and especially for those < 50 years old) are so small as to be hard to read: telling 1/1k vs 1/100k in a Sankey diagram doesn't work. They're both 1 pixel wide.
tomrod 5y ago
Sankey doesn't work well because we moved from SIR to SEIR+S with Delta -- meaning that we had breakthrough infections and the potential for waning immunity. Omnicron continues this. [0] https://en.wikipedia.org/wiki/Compartmental_models_in_epidemiology#The_SEIS_model https://en.wikipedia.org/wiki/Compartmental_models_in_epidem...

alea_iacta_est 5y ago

2 years in, we don't even have the numbers of people hospitalized from covid or with covid. It's not science reportings that is dead, it's science itself.

nradov 5y ago
Currently about half of COVID-19 patients in hospitals are incidental cases who were admitted for another reason and then tested positive during admission screening. https://gothamist.com/news/new-preliminary-state-data-shows-half-people-nyc-hospitals-covid-19-were-admitted-other-reasons https://gothamist.com/news/new-preliminary-state-data-shows-...
tomrod 5y ago
There are many people who have these numbers and study them. Public health officials in the US have both the reported numbers as well as piggy backing on the system set up for influenza monitoring (CLI & ILI). Simply because not all public reporting is as rich as, say, Illinois or Pennsylvania's, doesn't mean these figures aren't carefully scrutinized -- especially when threatened by political leaning from either side of aisle (such as when the then-current administration pressured the CDC to change and alter reports). In my experience, hospital operations officers, health information exchanges, and, depending on the state, state HHS agencies don't mess around. [0] https://www.cnbc.com/2021/04/09/trump-officials-bragged-about-pressuring-cdc-to-alter-covid-reports-emails-reveal-.html https://www.cnbc.com/2021/04/09/trump-officials-bragged-abou...