Thread by @MikeOtsuka

🚨If university students aren't tested every 2-3 days, "colleges are very likely to fall prey to outbreaks that will place vulnerable people on campus & in the surrounding community at risk for serious illness and death." 1/
consumer.healthday.com/infectious-dis…

This according to @ADPaltiel, a Professor of health policy at @YaleSPH, in comments on epidemological modelling, for which he was lead investigator, published today in JAMA👇. 2/
jamanetwork.com/journals/jaman…

See further comments👇by @ADPaltiel in above-linked press account. ("school" = "university") 3/

These findings are especially problematic, given these remarks of England's chief medical officer👇. If @UniversitiesUK re-open to in-person teaching without 2-3 day testing, other things will need to close to keep the pandemic under control. 4/4
theguardian.com/world/2020/jul…

PS: The study published in JAMA is co-authored by 2 Harvard scientists & has informed the university's decision to test students every 3 days as a condition of opening their halls of residence to reduced capacity re-opening. See thread👇. 1/

https://twitter.com/MikeOtsuka/status/1280568912573333504

Note that Harvard Arts & Sciences doesn't even regard testing every 3 days as sufficient to ensure the safety of in-person teaching & will be conducting their teaching entirely online in 2020-21. 2/

One further comment on Harvard's approach👇. 3/3

https://twitter.com/MikeOtsuka/status/1269362108422029317

Further comment by @ADPaltiel: Acknowledges challenge & expense of testing every 3 days but says that, given risks to faculty & staff, any university unwilling to test so frequently "has to ask itself if it has any business reopening". 1/
news.yale.edu/2020/07/31/stu…

Disappointing that @UniversitiesUK employers have not explained why, in spite of these findings, they nevertheless think it possible safely to re-open campuses to in-person teaching in the absence of such frequent testing. They simply ignore this challenge. @jim_dickinson 2/2

Here I go a bit into the weeds of @ADPaltiel's modelling. JAMA paper says "This model did not identify symptom-based screening alone as sufficient to contain an outbreak under any of the scenarios we considered." See also👇linked comments to the press. 1/
consumer.healthday.com/infectious-dis…

As @ADPaltiel also notes👆: "The reason [why testing of only those with symptoms is insufficient] is clear -- this virus can be transmitted by highly infectious asymptomatic silent spreaders". 2/

On the basis of a number of cited papers, the model assumes "a 30% probability that infection would eventually lead to observable COVID-19–defining symptoms in this young cohort" -- i.e., 70% of those 18-30 years old who have Covid-19 infection never develop symptoms. 3/

JAMA also includes a response👇to this paper, co-authored by @ehbvassar, a health economist who is President of Vassar College & past founder & Director of Yale Global Health Leadership Institute. 4/
jamanetwork.com/journals/jaman…

.@ehbvassar doesn't dispute @ADPaltiel's assertion that symptomatic testing is insufficient. But she argues that, for a small residential liberal arts college such as Vassar in a college town, testing all students (including asymptomatic) only every 4 weeks might be enough. 5/

What accounts for this difference? I think it's mainly that @ehbvassar believes Vassar College could improve on even the most optimistic of @ADPaltiel's three modelled scenarios (base, worst case, & best case). 6/

.@ADPaltiel's best case scenario assumes 10 new infections per week (from outside the student community) per 10,000 students. It also assumes an Rt of 1.5 (i.e., each infected student transmits to 1.5 others on average). 7/

By contrast, @ehbvassar assumes 4 new infections per week per 10,000 students "given that [Vassar students mainly live in halls of residence on campus and] we will limit student movement off campus and significantly reduce student interactions with faculty and employees". 8/

.@ehbvassar also assumes an Rt of 1.25 "given that we will have extensive social distancing, masking, and other measures". These measures are spelled out in more detail👇. Note that they don't much address transmission among students in halls of residence & while socialising. 9/

Presumably, students will be allowed to venture off campus in the evening. Assuming that the local bars are allowed to remain open, the risks of super-spreader events such as this👇will remain, even w social distancing. 10/

https://twitter.com/MikeOtsuka/status/1284758183521726464

Even if the bars of Poughkeepsie, New York, are all shut down, student will mix & socialise. There will be house parties👇. It's also possible that some will sleep with one another. @ehbvassar will not be able to impose & enforce monastic vows. 11/

https://twitter.com/MikeOtsuka/status/1285814841169584133

For the reasons spelled out👆, & in this thread👇, an Rt as low as 1.25 for students in residence stretches credulity. @ADPaltiel's best case scenario of Rt of 1.5 is more credible. (Note that "dorm" is American for UK "hall of residence".) 12/

https://twitter.com/gregggonsalves/status/1278368955044782081

What about @ehbvassar's assumption of only 4 new infections per 10k per week from outside the student community, rather than @ADPaltiel's best case of 10 per week? 13/

.@ADPaltiel calls these new infections "exogenous shocks" & describes how they might arise👇. 14/

The above-mentioned super-spreader event in the East Lansing bar is one good example of an "exogenous shock". It's not clear how @ehbvassar will manage to "limit student movement off campus" to prevent this. 15/

The number of new infections per week from outside the student population will also crucially depend on the general rate of infections in the surrounding community. 16/

In England, this has been rising in recent weeks as things have re-opened. We can infer from the latest ONS figures that the general rate in England is now about 5.5 new infections per week per 10,000 people. 17/
ons.gov.uk/peoplepopulati…

According to the ONS, this rate has also been almost twice as high among those 18-49 years old, in comparison with other age groups. 18/

For further evidence relevant to above modelling assumptions regarding rate of infection among university-aged, see this graph👇of prevalence of confirmed infections in Spain among ages 15-29, since easing of lockdown. 19/
bloomberg.com/opinion/articl…

Absent, therefore, significant reduction in general rate of infection in England this autumn -- contrary to prediction colder weather & school re-openings will lead to a rise -- it's unrealistic to assume better than @ADPaltiel's best case 10 per 10k weekly new infections. 20/

In support of @ADPaltiel's proposed & modelled frequent student testing, see this piece by @ashishkjha, Professor of Global Health at @HarvardHSPH, on why frequency of testing & rapidity of results is more important than accuracy of result per test. 21/
time.com/5873444/radica…

See also this paper by @michaelmina_lab (et al), to whom @ashishkjha refers, whose title says it all: "Test sensitivity is secondary to frequency and turnaround time for COVID-19 surveillance". 22/
medrxiv.org/content/10.110…

See also this brief thread👇by @nblqbl explaining why frequent & rapid, even if not very accurate, testing makes good sense. 23/

https://twitter.com/nblqbl/status/1290227482076028928

In a significant development, it appears that it will soon be possible for @UniversitiesUK employers to provide on-site, frequent, non-intrusive testing, with rapid results, for students & staff. See today's government announcement👇. 24/
gov.uk/government/new…

Further information on this morning's @BBCr4today👇:
▶️Interview w John Bell, Oxford’s Regius Professor of Medicine, starting at 7:13 am.
▶️Piece starting at 6:10 am.
▶️Interview with a government minister from 8:22 am. 25/
bbc.co.uk/programmes/m00…

Information gleaned from👆:
✅The test can be based on saliva (either throat swab or spit into a cup) rather than somewhat intrusive & unpleasant nasal swab.
✅It can test for the flu as well as Covid-19. (2🐦, 1🪨) 26/

✅Results within 90 minutes, through either a desktop machine or a palm-sized device which doesn't require technical expertise to use. So any university can set up rapid testing sites for students & staff. 27/27