I'm sitting here wondering what is going on with AZ #COVID__19 Hospital admissions/discharge data?

As of today, cumulative number of AZ Covid admissions today stands at 5272..

@wyattsheepie @ianmSC @andybiggs4az @AZDHS @Glenn_somebody_

That means that, over the course of the COVID pandemic, a total of 5272 have been hospitalized.

Total number of cumulative discharges?

12.8K

5 thousand Covid + patients have gone in.

12 thousand have come out.

Now, AZDHS notes that 63% (68K) of cases are missing hospitalization data.

However, http://azcentral.com  explains that the data lacking hospitlization status is limited to the demographic fields listed just below on the same page.

And that makes sense because the cumulative total from daily running admission numbers is 5430.

While not exactly the same as the published cumulative admission totals (5272), it's close enough (w/i 3%) to rule out any major data holes in hospitalizaion.

Sooo....

What we are looking at is roughly 5300 people admitted to AZ hospitals with confirmed or suspected COVID, and 12,803 discharged with confirmed or suspected COVID.

That means 2.4 times more people left the hospital with COVID than went in with it.

Now, despite there being an Executive Order (2020-37) requiring hosptials to report COVID admits and discharges within 24 hours, AZ central cites a 7-10 day reporting lag, which we do see in the most recent days, at least for admissions.... https://azgovernor.gov/executive-orders

If we credit the 10 day reporting lag to explain the discrepancy, there would have to be an average of roughly 750 more COVID admissions per day during those incomplete days to have admitted as many people entering the hospital with a confirmed or suspected Covid as leaving...

750 in one day is 7 TIMES more admissions in a single day than the highest number AZ has seen so far (106 on June 15th.)

And that seven-fold increase would had to occurred for the full 10 day lag.

That's not possible. Sooo....

What might be going on? I've thought about it and asked a lot of smart people to opine.

Possibilities?

#1) Nocosomial infections
#2) Asymptomatic positives from non-COVID patients
#3) Data collection flaws
#4) Data manipulation
#5) Other possibilities

There is definitely some of #1 and #2.

Nocosomial (hospital-aquired) infections have been a problem with COVID and, given the vulnerabilities of hospital populations, it's not too surprising. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7180328/

There's probably less of #2 than you might think since patients require a negative test before elective procedures.

Certainly pregnant women & emergent (non-COVID) patients could fall into this group, as well as those who aquire COVID between testing & admission...but not 7500.

Az doesn't release raw patient data so I couldn't dig too deep..but Ohio does.

In Ohio, there are multiple incidences of patients with onset dates which predate their COVID admissions by weeks/months...

https://coronavirus.ohio.gov/wps/portal/gov/covid-19/dashboards/overview

...which brings both #3 and #4 into focus.

They both seems plausible. Maybe there's double-counting somewhere. Idk.

But it certainly seems that the admissions numbers line up reasonably with the current census and mandatory reporting. The discharges? Not so much.