Read on Twitter
@inschool4life, Nathaniel Hupert and I recently crunched some numbers suggesting millions of undetected COVID cases in the US the month of March.
What have we learned since then, what do we still need to know? Is it likely millions in US have had COVID?
https://www.medrxiv.org/content/10.1101/2020.04.01.20050542v2
What have we learned since then, what do we still need to know? Is it likely millions in US have had COVID?
https://www.medrxiv.org/content/10.1101/2020.04.01.20050542v2
One study found 33/215 or 15.3% of women in labor in NYC from March 22 - April 4 had COVID.
If they were representative of NYC, that would be 1.3 million cases, just in NYC. Unclear, however, if LD is representative (nosocomial infns, inf-induced labor) https://www.nejm.org/doi/full/10.1056/NEJMc2009316
If they were representative of NYC, that would be 1.3 million cases, just in NYC. Unclear, however, if LD is representative (nosocomial infns, inf-induced labor) https://www.nejm.org/doi/full/10.1056/NEJMc2009316
March 18 - 25, SARS-CoV-2 was found in sewage in MA. Going from sewage to prevalence is tenuous, but of note, their viral titers double every ~3 days consistent with the growth rate of US deaths.
https://www.medrxiv.org/content/10.1101/2020.04.05.20051540v1.full.pdf
https://www.medrxiv.org/content/10.1101/2020.04.05.20051540v1.full.pdf
San Miguel Cnty, CO produced 4/1 found <1% seropositive but an additional 2% showing an increased chance of becoming positive. In county of 8,200, this is 82-246 people infected at a time when they had 1 confirmed case, in line w/ our 1/100 case-detection
https://www.sanmiguelcountyco.gov/CivicAlerts.aspx?AID=511
https://www.sanmiguelcountyco.gov/CivicAlerts.aspx?AID=511
We've found others using ILI in other countries, such as this group in France, and following a similar logical series as our analysis - find excess ILI, see it correlates w/ COVID across regions, and discuss changes in care-seeking.
https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2020.25.14.2000326
https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2020.25.14.2000326
While Gangelt showed 14% seroprevalence and Scotland seroprevalence low but on the rise, these numbers are harder to cross-check with our US estimates because we don't have ILI nor well-defined epidemic start-dates for those places as were used to estimate subclinical rates.
36% of people at a homeless shelter were PCR-positive, but they aren't a representative sample of MA. A good study and some important questions arising from it, just inconclusive w.r.t. our ILI analysis. https://www.medrxiv.org/content/10.1101/2020.04.12.20059618v1
An exciting lead is this paper proposing an alternative view of COVID as an intestinal pathogen w/ respiratory complications, and why we may see non-ILI (e.g. GI) presentations & mechanism of subclinical/asymptomatic infections. We're looking into it!
https://www.biorxiv.org/content/10.1101/2020.04.06.028712v2.full.pdf
https://www.biorxiv.org/content/10.1101/2020.04.06.028712v2.full.pdf
This paper adds to questions about Se of NP swabs in general population. We recommend serology + multi-swab (NP + rectal/stool) surveys with date-stamps to ensure we don't miss any cases and can properly assess trends in prevalence as an additional test of our hypotheses.
For example, if Gangelt sampled 3-days later and found 28% seroprevalence, it would suggest some combo of the epidemic 2-3 weeks prior was doubling every 2-3 days and a smaller false-positive rate of the serology test. Same for LD swabs - did LD prevalence increase over March?
On the flip side, if Gangelt sampled 2 weeks later and found lower (e.g. 10%) seroprevalence, we would question how much such serosurveys actually told us. In general, disaggregating serology and swabs by date-x-location dramatically improves the value of those samples.
We'll keep looking for examples of swabs & serology to see what's right/wrong about our analysis.
Somewhere out there is the truth about COVID - that's what we're after. Please keep sharing papers you find, especially those disagreeing with our analysis - we're on team science!
Somewhere out there is the truth about COVID - that's what we're after. Please keep sharing papers you find, especially those disagreeing with our analysis - we're on team science!
Special thanks to @Nateum_s @LourencoJML @DanRosenheck @SethS_D @bolkerb @jd_mathbio @NahasNewman and more for thinking continuously & critically about these questions.
There are many smart people out there whose thoughts add value and help the world get closer to the answers!
There are many smart people out there whose thoughts add value and help the world get closer to the answers!
