Long overdue 🧵 on a COVID study we preprinted last year, now out in JID. Co-led by @nialljlennon @gabriel_stacey @eric_lander, enabled by @BroadGenomics & @MassDPH
Key finding: shed viral load is independent of symptoms at time of testing (pre-vax)
academic.oup.com/jid/advance-ar…
1/n
Key finding: shed viral load is independent of symptoms at time of testing (pre-vax)
academic.oup.com/jid/advance-ar…
1/n
2/14
Back in Apr 2020, COVID overwhelmed our hospital in MA (I was on service & will never forget).
It also swept thru nursing homes, so @MassDPH tested ALL residents+staff of SNFs for COVID (back when tests were rare: big job w help from MA Nat'l Guard).
Back in Apr 2020, COVID overwhelmed our hospital in MA (I was on service & will never forget).
https://twitter.com/roby_bhatt/status/1330577650310651904
It also swept thru nursing homes, so @MassDPH tested ALL residents+staff of SNFs for COVID (back when tests were rare: big job w help from MA Nat'l Guard).
3/14
B/c of @broadinstitute's remarkable early scale-up of testing, >70% of these (>30,000 over 2+ mo) were run @BroadGenomics. @nialljlennon & @gabriel_stacey recognized an opportunity to study viral loads on their standardized platform; @MassDPH had queried symptoms (yes/no).
B/c of @broadinstitute's remarkable early scale-up of testing, >70% of these (>30,000 over 2+ mo) were run @BroadGenomics. @nialljlennon & @gabriel_stacey recognized an opportunity to study viral loads on their standardized platform; @MassDPH had queried symptoms (yes/no).
4/14
So, we could ask how a binary assessment of symptom status at the time of testing correlated w/ viral load.
We found no relationship at all.
People without symptoms were just as likely to be shedding 1 billion copies per mL of viral RNA as people with them.
So, we could ask how a binary assessment of symptom status at the time of testing correlated w/ viral load.
We found no relationship at all.
People without symptoms were just as likely to be shedding 1 billion copies per mL of viral RNA as people with them.
5/14
That is the main takeaway from our paper: even w/o symptoms, ppl can shed tons of virus. This was pre-vaccines, pre-Delta or Alpha. But already we knew that asymptomatic or presymptomatic transmission was a huge problem: one of the biggest challenges to containing COVID imo.
That is the main takeaway from our paper: even w/o symptoms, ppl can shed tons of virus. This was pre-vaccines, pre-Delta or Alpha. But already we knew that asymptomatic or presymptomatic transmission was a huge problem: one of the biggest challenges to containing COVID imo.
6/14
(Variability has been a huge challenge for COVID too… bad enough to kill millions & collapse health care systems; variable enough that it's way more common to know someone who had a mild case: everyone knows an Uncle Bob who's smoked for 50 yrs but did fine… but I digress)
(Variability has been a huge challenge for COVID too… bad enough to kill millions & collapse health care systems; variable enough that it's way more common to know someone who had a mild case: everyone knows an Uncle Bob who's smoked for 50 yrs but did fine… but I digress)
7/14
Anyway, the data held more surprises. We had heard anecdotally that Ct had been rising (VL dropping) over time as the initial pandemic wave receded. @AnnWoolleyMD helped by looking at BWH Ct data that confirmed this.
Anyway, the data held more surprises. We had heard anecdotally that Ct had been rising (VL dropping) over time as the initial pandemic wave receded. @AnnWoolleyMD helped by looking at BWH Ct data that confirmed this.
8/14
So we dug deeper. And while there was no difference in viral loads as the local epidemic peaked (Apr 2020), a difference emerged over time as it waned: later in our study (Jun 2020), those without symptoms had lower viral loads than those with symptoms.
So we dug deeper. And while there was no difference in viral loads as the local epidemic peaked (Apr 2020), a difference emerged over time as it waned: later in our study (Jun 2020), those without symptoms had lower viral loads than those with symptoms.
9/14
What gives? Why would viral load's relationship to symptoms change w month? Turns out it relates to epidemic dynamics, & to trajectories of viral shedding (peaks early, declines slowly) & symptoms (start early in illness, if they happen at all). I tried to diagram that here.
What gives? Why would viral load's relationship to symptoms change w month? Turns out it relates to epidemic dynamics, & to trajectories of viral shedding (peaks early, declines slowly) & symptoms (start early in illness, if they happen at all). I tried to diagram that here.
10/14
As case counts rise, most cases are recent, so regardless of symptoms, most folks have high VLs. Later, as cases recede, cases happened longer ago on avg. But those w symptoms are selected to be early in illness, when VLs are higher; not so for those w/o symptoms.
As case counts rise, most cases are recent, so regardless of symptoms, most folks have high VLs. Later, as cases recede, cases happened longer ago on avg. But those w symptoms are selected to be early in illness, when VLs are higher; not so for those w/o symptoms.
11/14
I again tried to diagram that here.
Sx = red, no Sx = red or blue
top = rising cases (early in study)
bottom = declining cases (later in study)
Those w/o symptoms can accumulate in the "clearance" phase, with low VL, causing the "pileup" seen at high Ct in the histogram.
I again tried to diagram that here.
Sx = red, no Sx = red or blue
top = rising cases (early in study)
bottom = declining cases (later in study)
Those w/o symptoms can accumulate in the "clearance" phase, with low VL, causing the "pileup" seen at high Ct in the histogram.
12/14
If this sounds familiar, maybe it's b/c co-author @michaelmina_lab proved it more rigorously w/ @jameshay218: the shape of a Ct histogram (from a random cross-section of people - this is key) can predict epidemic dynamics in a single measurement.
science.org/doi/10.1126/sc…
If this sounds familiar, maybe it's b/c co-author @michaelmina_lab proved it more rigorously w/ @jameshay218: the shape of a Ct histogram (from a random cross-section of people - this is key) can predict epidemic dynamics in a single measurement.
science.org/doi/10.1126/sc…
13/14
Caveats:
- no longitudinal f/u = can't tell asympt from presympt
- severe cases excluded (b/c they'd have been in hospital, not SNF)
- Ct ~ viral RNA copies (b/c of standard curve), but ≠ infectious virus; still, we think Ct informs risks of spread (esp pre-vaccine)
Caveats:
- no longitudinal f/u = can't tell asympt from presympt
- severe cases excluded (b/c they'd have been in hospital, not SNF)
- Ct ~ viral RNA copies (b/c of standard curve), but ≠ infectious virus; still, we think Ct informs risks of spread (esp pre-vaccine)
14/14
There's more nuance in the article, but I learned a lot from this study - it shaped how I think abt SARS-CoV-2 shedding & spread - so I thought I'd share.
Thx to co-authors tagged above + @HeidiRehm, Sandy Smole, & Deb Hung (tweetless), & JID editors @IDSAInfo @OUPMedicine
There's more nuance in the article, but I learned a lot from this study - it shaped how I think abt SARS-CoV-2 shedding & spread - so I thought I'd share.
Thx to co-authors tagged above + @HeidiRehm, Sandy Smole, & Deb Hung (tweetless), & JID editors @IDSAInfo @OUPMedicine
15/14
I'll add: the non-intuitive finding that epidemic phase affects measured VL affects analysis of variant & post-vax VLs.
New variants, by def'n, are increasing in freq as they emerge = measured VL will artifactually be higher (even if "true" VL trajectory unchanged).
I'll add: the non-intuitive finding that epidemic phase affects measured VL affects analysis of variant & post-vax VLs.
New variants, by def'n, are increasing in freq as they emerge = measured VL will artifactually be higher (even if "true" VL trajectory unchanged).
16/14
Similarly, since #VaccinesWork, infections become less frequent after vaccination; so measured VL will artifactually be lower (cases less likely to be recent).
Variants or vax may have TRUE effects on VL, aside from this artifact. We just have to be careful how we measure.
Similarly, since #VaccinesWork, infections become less frequent after vaccination; so measured VL will artifactually be lower (cases less likely to be recent).
Variants or vax may have TRUE effects on VL, aside from this artifact. We just have to be careful how we measure.
17/14
In assessing measured VL, timing of test (& thus the reason it was sought) is crucial for interpreting results.
Best case is full longitudinal trajectories! But that's hard to come by, unless you're @yhgrad studying the NBA:
jamanetwork.com/journals/jamai…
medrxiv.org/content/10.110…
In assessing measured VL, timing of test (& thus the reason it was sought) is crucial for interpreting results.
Best case is full longitudinal trajectories! But that's hard to come by, unless you're @yhgrad studying the NBA:
jamanetwork.com/journals/jamai…
medrxiv.org/content/10.110…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
