It’s been a week since this article by
@DrLeanaWen
ignited a firestorm of anger from all sides.
To some extent I think what is happening is a *false controversy* stemming from the fact that we are not speaking the same language
Definitions matter.🧵
11
11
71
Replies
Public health surveillance definitions exist so disease burden can be estimated. The definitions balance accuracy and the feasibility of data collection. As validation studies are done and feedback given (eg from those who submit the data), surveillance definitions are adjusted.
2
0
14
For example, hospitals track catheter associated bloodstream infection. Recently, CDC added exclusion criteria such that infection in a patient with certain other high-risk devices is excluded. This fixes overcounting.
1
0
11
We also track ventilator associated pneumonia. Recently CDC changed the definition entirely, since the earlier definition hinged on radiology reports which were onerous to review and too subjective.
How does this relate to COVID?
1
0
10
What is the official definition of a “COVID hospitalization”? When we say both “with and for” COVID are being counted in the official #️⃣s that doesn’t mean doctors are lying about diagnoses. The doctor’s diagnosis does not enter the picture when it comes to public reporting.
1
0
13
Here’s the definition (thanks
@KelleyKga
for the markup). Check out the bottom. A patient can have had COVID 4 months ago and still be a COVID hospitalization. I see this every day. There’s no nefarious intent here. But it’s time for an update. We should specify both timing…
2
1
29
…and severity.
With
@wbranchelliman
we showed that dexamethasone receipt tracks closely with other measures of severity.
1
0
19
That’s why dexamethasone is how MA tracks severity of disease. Dex isn’t “with versus for”. It’s dex. But it shows that severity is decreasing over time and that there’s a need for a better surveillance definition.
Proportion with dex started around 55% a year ago and now 30%.
1
0
16
At my hospital we are now having days where we dip below 10% of COVID+ patients on dex. That’s such good news.
Some of these are completely asymptomatic. Those are overcounted COVID hospitalizations.
1
0
22
This is all way more important in hospitals/states where universal asymptomatic admission screening is happening. That’s not everywhere. Not at all. And I expect that to decrease over the next weeks to months given the SHEA statement recommending against it.
1
0
17
Now to poke the 🐻 …what’s the official definition of a COVID death? There are several!
I previously linked to an older CSTE (state and territorial health dept) guideline. Sorry. But it’s not as important as you might think. It’s a guideline not a mandate…
2
0
11
States measure COVID deaths in whatever way is feasible for them.
My state uses positive test within 30 days of death OR COVID listed on death certificate.
Call yours and ask what they use.
State data feed the CDC daily tracker and other trackers.
1
0
11
On the other hand, the National Center for Health Statistics gets its data from death certificates only. That’s where the total COVID death toll number comes from.
Are death certificates a more accurate way to count COVID deaths? Probably.
1
0
10
Again, there’s no conspiracy theory here, but if you’ve ever filled out a death certificate (and, uh, yes, I have) or even read one after caring for a patient who died, you know how complicated it is to get it right. Humans are complex.
2
0
14
My biggest concern is how these #️⃣s are used to make policy, including vax recs. Overcounting hospitalizations and/or deaths will underestimate vax effectiveness. We showed this in a 2nd study.
3
0
13
So I agree with
@DrLeanaWen
:
“Hospitals and health departments should use a uniform set of criteria to classify covid hospitalizations and deaths.”
And let’s study this so we can “put to bed the criticism that severe illness was overcounted all along.”
5
0
19
@ShiraDoronMD
@DrLeanaWen
What was conceived as surveillance definitions absolutely filtered down to state/county guidance and impacted how death certificates were completed for this “novel disease.”
The difference between NNDSS and NVSS numbers isn’t that great.
Several examples:
Hospice Deaths by year.
pre-2020 the UCOD was one of these 20 respiratory causes ~ 0.10 % of the time.
2020-2022 Covid was the UCOD in Hospice ~ 4.5% of the time.
But sure; the idea we've been overcounting has been "debunked", right
@mehdirhasan
@DrLeanaWen
2
6
13
1
0
0
@ShiraDoronMD
@DrLeanaWen
I would think people would be happy to hear Covid was not the leading cause of death.
0
0
1
@ShiraDoronMD
@DrLeanaWen
Excellent take Shira! Thank god there is a voice of reason in the sea of epidemiology.
0
0
1
@ShiraDoronMD
@DrLeanaWen
Definitions DO matter.
Dexamethasone use is one criteria for hospitalization due to COVID & addresses only respiratory disease. Excess deaths is used globally.
A major argument with the oped was that no data was presented to support a strident minimizing statement.
#WeCanDoBetter
0
0
2
@ShiraDoronMD
@DrLeanaWen
Just a reminder that this 'different language' was effectively banned for much of the last two years. And there are many who would like to see any discussion quashed even now - as the reaction to Wen's article makes clear.
0
0
0