Read on Twitter
/ #COVID19 hospitalizations are on the rise, as visualized by @awitherspoon, and contextualized by @sam_baker 
Here, I’ll break down hospitalization data, and explain why we must avoid drawing conclusions or drafting policy based on these data *alone* https://www.axios.com/coronavirus-hospitalizations-increasing-abc7e1f7-51b1-4b5c-a2e8-ab55685ac522.html?utm_campaign=organic&utm_medium=socialshare&utm_source=twitter
The first thing to note about hospitalization data is that both the numerator and denominator change on a day-to-day basis. Let’s start by looking at the numerator:
Most #COVID19 hospitalizations occur 8 - 12 days after the onset of symptoms
src: https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance-management-patients.html
Most #COVID19 hospitalizations occur 8 - 12 days after the onset of symptomssrc: https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance-management-patients.html
The typical length of stay for hospitalized #COVID19 patients is 10 to 13 days...BUT, this range is based on data from China reported early in the pandemic.The treatment of COVID-19 has changed over time.
src: https://www.cdc.gov/coronavirus/2019-ncov/hcp/clinical-guidance-management-patients.html
There are now two treatments that are proven to be helpful in managing #COVID19 in hospitalized patients:
Remdesivir has been shown to reduce the typical (median) amount of time patients spend in the hospital from 15 
10 days
ref: https://www.nejm.org/doi/full/10.1056/NEJMoa2007764
Remdesivir has been shown to reduce the typical (median) amount of time patients spend in the hospital from 15 10 daysref: https://www.nejm.org/doi/full/10.1056/NEJMoa2007764
Dexamethasone was shown to decreased the amount of time intubated #COVID19 patients spent on a mechanical ventilator, but there was also a small (indirect) shortening of hospital length of stay (median 12 vs. 13 days)ref: https://www.nejm.org/doi/full/10.1056/NEJMoa2021436
ref: https://www.acc.org/latest-in-cardiology/journal-scans/2020/08/03/10/18/dexamethasone-in-hospitalized-patients
Taken together, we expect #COVID19 hospitalizations to be shorter today, than they were in March through June. And that’s exactly what we’re seeing. In Colorado, for example:
ventilator use
length of stay
H/T @CODaleyNews ( https://www.cpr.org/2020/10/16/new-hospital-data-shows-ventilator-use-fell-over-time-and-coronavirus-patient-stays-grew-shorter/)
src: @COHospitalAssn
ventilator use length of stayH/T @CODaleyNews ( https://www.cpr.org/2020/10/16/new-hospital-data-shows-ventilator-use-fell-over-time-and-coronavirus-patient-stays-grew-shorter/)
src: @COHospitalAssn
So, when looking at the number of #COVID19 hospitalizations over time, we must bear in mind that the number of hospitalizations on any given day (numerator) should be lower today, than it was over the summer — *assuming nothing about the disease or the patients has changed.*
I need to step away for a bit, but the upshot so far is:
Shorter #COVID19 hospitalizations translate into fewer people in the hospital for #COVID19 on any given day *so long as community transmission remains the same*
Scientists have found ways to shorten hospitalizations
Shorter #COVID19 hospitalizations translate into fewer people in the hospital for #COVID19 on any given day *so long as community transmission remains the same*Scientists have found ways to shorten hospitalizations
Shifting gears to the number of hospital beds (the denominator): This value*IS NOT* about bed capacity per se. This value says:
A physical bed exists
A medical team is available to care for the patient
A physical bed existsA medical team is available to care for the patient
The number of beds available tells us the number of patients that can be *safely* admitted to the hospital (for any reason)
Similarly the number of ICU beds available tells us the number of patients that can be *safely* admitted to the ICU (for any reason)
Key data is missing..
Similarly the number of ICU beds available tells us the number of patients that can be *safely* admitted to the ICU (for any reason)
Key data is missing..
The # does *not* capture whether:
the bed is in a unit or ward that can accommodate #COVID19 patients
the *available* staff has the training to safely care for COVID patients
adequate PPE is available
These factors *actual* bed available
CC @meganranney @Cleavon_MD
the bed is in a unit or ward that can accommodate #COVID19 patientsthe *available* staff has the training to safely care for COVID patientsadequate PPE is availableThese factors
*actual* bed availableCC @meganranney @Cleavon_MD
These factors are most relevant for non-ICU patients, but that’s where “swing” beds come into play— swing beds are one reason why the denominator can change.
There are two definitions of “swing” beds — here we use it to mean regular beds that can be made into “ICU-level” beds.
There are two definitions of “swing” beds — here we use it to mean regular beds that can be made into “ICU-level” beds.
There are also physical beds scattered throughout a hospital that aren’t technically meant to be assigned to a patient that will be staying overnight — like an emergency room.
Taken together, these “extra” beds comprise a hospital’s “surge capacity” — they
*actual* beds
Taken together, these “extra” beds comprise a hospital’s “surge capacity” — they
*actual* beds
To recap — there are multiple factors that and *actual* hospital bed capacity. These factors affect every level of care, from the emergency room all the way to the ICU.
@TXMedCenter does an exceptional job of breaking down *actual* ICU numbers
see: https://www.tmc.edu/coronavirus-updates/overview-of-tmc-icu-bed-capacity-and-occupancy/
and *actual* hospital bed capacity. These factors affect every level of care, from the emergency room all the way to the ICU.@TXMedCenter does an exceptional job of breaking down *actual* ICU numbers
see: https://www.tmc.edu/coronavirus-updates/overview-of-tmc-icu-bed-capacity-and-occupancy/
Best I can tell, neither state agencies nor HHS is consistently reporting *actual* #COVID19 bed availability.
When I was still working closely w/ FEMA, they were using @DefinitiveHC’s dataset, but key details are opaque since HHS changed things in July: https://www.aha.org/news/headline/2020-08-03-hhs-updates-covid-19-guidance-and-faqs-hospitals-daily-data-reporting
When I was still working closely w/ FEMA, they were using @DefinitiveHC’s dataset, but key details are opaque since HHS changed things in July: https://www.aha.org/news/headline/2020-08-03-hhs-updates-covid-19-guidance-and-faqs-hospitals-daily-data-reporting
What we *do know* is that hospitals are reporting these key details to HHS via their new system.
We know that based on the guidance document issued by HHS to hospitals: https://www.hhs.gov/sites/default/files/covid-19-faqs-hospitals-hospital-laboratory-acute-care-facility-data-reporting.pdf
We know that based on the guidance document issued by HHS to hospitals: https://www.hhs.gov/sites/default/files/covid-19-faqs-hospitals-hospital-laboratory-acute-care-facility-data-reporting.pdf
We also know that HHS is collecting information about the availability of PPE on a weekly basis, and the information they collect is remarkably specific — including whether the hospital anticipates a shortage, whether staff reuse PPE, and more.
What we don’t know is what @HHSGov does with that information— and why key details are not made available to the public.
To date, HHS has limited public datasets to an *estimated* hospitalization count at the state level.
https://protect-public.hhs.gov/datasets/state-representative-estimates-for-hospital-utilization
To date, HHS has limited public datasets to an *estimated* hospitalization count at the state level.
https://protect-public.hhs.gov/datasets/state-representative-estimates-for-hospital-utilization
In summary, drawing conclusions or changing policy based on #COVID19 hospitalization data is fraught with caveats, data gaps, and pitfalls.
We further established that HHS collects the data we need but fails to make that data publicly available…
We further established that HHS collects the data we need but fails to make that data publicly available…
Final thoughts:
I fail to see why @HHSGov desires to keep the American people in the dark vis Ă vis real-time hospitalization data (there are easy ways to make the data available without compromising patient privacy)
PPE inventories should also be public
/end
CC @getusppe
I fail to see why @HHSGov desires to keep the American people in the dark vis Ă vis real-time hospitalization data (there are easy ways to make the data available without compromising patient privacy)PPE inventories should also be public/end
CC @getusppe
Examples of why this matters:
Hospital data transparency sheds light on interdependencies, helps 
scarce resources
Right now we’re not getting the full picture, the data can’t tell a complete story
By the time we’re aware of a problem, it’s already too late
H/T @MollyBeck https://twitter.com/wihealthnews/status/1318238701877710848
Hospital data transparency sheds light on interdependencies, helps scarce resourcesRight now we’re not getting the full picture, the data can’t tell a complete storyBy the time we’re aware of a problem, it’s already too lateH/T @MollyBeck https://twitter.com/wihealthnews/status/1318238701877710848
