An analysis of the results of routine employee testing for SARS-like infections within the WIV and other Wuhan labs raises serious issues about their validity

Abstract

Various representations were offered during the China-WHO joint-team visit about the routine testing of employees at some Wuhan labs. The final China-WHO joint-team report used these representations as an argument to dismiss the possibility of the COVID-19 outbreak being caused by an early infection in a Wuhan institution involved in BatCov research.

A review of the test results mentioned in the China-WHO report and in interviews shows multiple issues both with the relevance of the tests, in terms of the exact institutions covered and the populations within these institutions, but also with the purported results of these routine tests which at best seem to point to very limited number of employees being tested.

Full text

An analysis of the results of routine employee testing for
SARS-like infections within the WIV and other Wuhan labs
raises serious issues about their validity
Authors: Steven Quay1(MD, PhD), ORCID; Gilles Demaneuf2(MS Eng., Data Scientist), ORCID
1Email: Steven@DrQuay.com
2correspondence to: gilles@demaneuf.com
Abstract:
Various representations were offered during the China-WHO joint-team visit about the routine testing of
employees at some Wuhan labs. The final China-WHO joint-team report used these representations as
an argument to dismiss the possibility of the COVID-19 outbreak being caused by an early infection in
a Wuhan institution involved in BatCov research.
A review of the test results mentioned in the China-WHO report and in interviews shows multiple issues
both with the relevance of the tests, in terms of the exact institutions covered and the populations within
these institutions, but also with the purported results of these routine tests which at best seem to point
to very limited number of employees being tested.
Main Findings:
1. The tests did not cover all relevant institutions
Some key institutions that are attested to have worked on bat coronaviruses (BatCoV) at BSL2 or BSL3
are not mentioned at all in the statements.
2. The tests performed in one of the institutions mentioned are irrelevant
The tests done at the Jianghan CDC cannot establish if some laboratory staff got infected in late 2019,
as these tests were performed in June 2020 and were PCR tests which can only detect a positive
infection within a few weeks of the first symptoms. In any case the Jianghan CDC itself is most likely
irrelevant anyway as not being known for any BatCoV research. The mention of the Jianghan CDC in
the Annex D5 of the China-WHO joint-team report can thus only create confusion.
3. The tests did not cover all relevant people
For a proper detection of possible research-related infections in 2019 that could have sparked the
pandemic, the correct approach would have been to test (1) all field sampling staff, (2) all staff, students
and temporary workers at institutions handling coronaviruses, (3) a sample of people living in proximity
(especially downwind) from all these institutions.
4. A proper testing at the WIV would have returned positive cases
Given the prevalence of COVID-19 in Wuhan during the first half of 2020, with near absolute certainty
some of the WIV 590 staff and students would have tested positive (igG+) in March 2020, following an
infection with SARS-CoV-2.
Hence the statement that nobody at the WIV tested positive for SARS-CoV-2 is either (a) misleading as
being based on an unreliably small fraction of the entire WIV population that should be tested, or (b) it
is simply untrue.
The situation is made worse when considering the four labs mentioned by Peter Embarek.
5. Clarification of the test populations and results is urgently required
The WIV and other Wuhan labs of interest should either (a) clarify that they only did limited testings and
detail exactly who was tested and when, or (b) provide case histories of the employees/students that
necessarily did test positive if they tested large enough populations.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 1/20

1. Background:
Since the early days of the COVID-19 pandemic, the possibility of the COVID-19 pandemic being the
result of a lab-related accident has been regularly mentioned and has triggered various polemics. On
their side, Chinese officials tried to address these concerns by pointing to the results of routine health
checks of labs personnel, including antibodies testing for SARS-Like infections - which they claimed
were all negatives.
In January 2021 the departing US Secretary of State nevertheless issued a factsheet, which while
stating clearly that the the US Government had reached no specific conclusion, pointed to possible
early infections involving lab personnel, without disclosing the source:
"We have not determined whether the outbreak began through contact with infected animals or was
the result of an accident at a laboratory in Wuhan, China.
[--]
The United States government has reason to believe that several researchers inside the WIV
became sick in autumn 2019, before the first identified case of the outbreak, with symptoms
consistent with both COVID-19 and common seasonal illnesses. This raises questions about the
credibility of WIV senior researcher Shi Zhengli’spublic claim that there was “zero infection” among
the WIV’s staff and students of SARS-CoV-2 or SARS-related viruses.” (source: fact-sheet)
In April 2021 the China-WHO joint-team published its final report, which considered as extremely
unlikely the possibility of COVID-19 being the result of a lab-related accident. The negative routine
testings in some Wuhan labs (including the WIV) were a key element in the joint-team assessment
“The three laboratories in Wuhan working with either CoVs diagnostics and/or CoVs isolation and
vaccine development all had high quality biosafety level (BSL3 or 4) facilities that were
well-managed, with a staff health monitoring programme with no reporting of COVID-19 compatible
respiratory illness during the weeks/months prior to December 2019, and no serological evidence
of infection in workers through SARS-CoV-2-specific serology-screening.” (source: China-WHO
joint-team report, page 119).
Given the importance of these routine testings in the origins debate, adetailed and careful evaluation
of the significance of these testings and of their purported results is required.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 2/20

2. Available statements about tests:
We searched the China-WHO joint-team report and various interviews with the media for the available
joint-team statements about the serological tests for SARS-CoV-2 and found four key statements listed
below.
As can be seen these statements are not particularly specific on the exact population tested. In
particular we could have reasonably expected the Annexes D5 and D7 of the China-WHO report to be
much more precise on this subject, given its crucial importance:
S1:Shi Zhengli’s Reply to Science Magazine, 27th July 2020:
Q: Is it possible that someone associated with the institute became infected in some other way,
for instance while collecting, sampling, or handling bats?
A: Such a possibility did not exist. Recently we tested the sera from all staff and students in the
lab and nobody is infected by either bat SARSr-CoV or SARS-CoV-2. To date, there is "zero
infection" of all staff and students in our institute.
S2:Annex D7 of the China-WHO report on Covid-19 origins:
●The reserved sera in April 2019 and March 2020 from all the workers and students in the
research group led by Professor Shi Zhengli were seronegative for SARS-CoV-2 antibodies.
S3:Annex D5 of the China-WHO report on Covid-19 origins:
●Hubei [region] CDC. All its laboratory staff have been tested for SARS-CoV-2-specific
antibodies: all had negative IgM and IgG results.
● Wuhan [city] CDC. one of its staff was confirmed SARS-CoV-2 seropositive after infection due
to family cluster transmission. All other staff have tested negative. A health check is mandatory
for all BSL-2 laboratory workers, but no serum is preserved.
● Jianghan [district] CDC. All PCR tests for SARS-CoV-2 of all laboratory workers in June 2020
were negative.
S4:Peter Embarek quoted in Science Magazine,14th Feb 2021:
Q: But my question is whether you learned anything new in China. Now that you’ve been there,
do you have more reason to say it’s “extremely unlikely” than before?
A: Yes. We had long meetings with the staff of the Wuhan Institute of Virology and three other
laboratories in Wuhan. They talked about these claims openly. We discussed: [--] ‘Did you test
your staff? [--]. They had retrospectively tested serum from their staff. They tested samples
from early 2019 and from 2020.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 3/20

3. Who should have been tested?
The population to test for the purpose of detecting a lab-related infection depends on the lab-related
accident scenario considered. The table below shows the candidate populations for the 3 main
scenarios: field sampling accident, Lab Acquired Infection (LAI) and Lab escape without LAI:
Lab-Related
Accident
Scenario:
Field sampling accident
Lab acquired infection (LAI)
of Wuhan lab personnel
Lab escape without LAI
LS1
LS2
LS3
Description
Infection during field
sampling by or on behalf of a
Wuhan laboratory
Infection inside a Wuhan institution
with laboratories
Infection outside a Wuhan
institution with laboratories
Index case
Personnel present at field
sampling site, went back to
Wuhan or infected someone
who went back to Wuhan1
Can be lab personnel, staff, student,
or anybody present in the institution
(including temporary worker or
visitor).
Someone out of the lab, in
proximity typically, or in relation
to lab activities (such as waste
processing)
Actual Biosafety
Level
Very limited to equivalent of
Biosafety Level 2 (BSL-2)
laboratory, full PPE
commonly not worn2
Bat SARS-related coronaviruses
research officially performed at
BSL-2 & BSL-3 levels
Bat SARS-related coronaviruses
research officially performed at
BSL-2 & BSL-3 levels
Incident
Either via contact with
animal hosts or animal waste
on site
● infection in lab suite handling
virus
● infection in common facilities
shared with lab suite handling
virus
● infection in institution precinct via
aerosols, wastes or stray lab
animal
● infection via aerosol outside
lab precinct
● infection via incompletely
neutralized liquid or solid lab
wastes outside lab
● infection via stray lab animal
First person
infected in a
lab/institution?
Possibly - could also be
employee, contractor,
contract worker, collaborator,
associate, or visitor,
university student
Probably - could be employee,
contractor, contract worker,
collaborator, associate, or visitor
First person is infected outside
of lab/institution
Entity of Interest
Any field sampling team
Any institution in Wuhan working on
BatCoVs (may include Wuhan
Institute of Biological Products
(WIBP))
Any institution in Wuhan working
on BatCoVs (may include WIBP)
Who should be
tested?
Anybody present during field
sampling
●Direct Infection in lab suites: All
people with access to the lab
suites and lab wastes.
●Indirect Infection via contact:
Anybody sharing space/facilities
with people working directly in lab
suites or handling the lab wastes
inside the institution.
●Indirect infection via aerosols:
Anybody within the lab institution
●Indirect Infection via
contact: Anybody in proximity
to any liquid wastes from the
institution or handling the lab
wastes outside of the
institution.
●Indirect infection via
aerosols: Anybody in
downwind proximity of the
institution.
Table 1:Llab-related accident scenario and corresponding candidate population for an index-case
2It has been amply proven that PPEs are often not worn during field specimen collection trips. For instance
Shi Zhengli herself explained that most of the time only ordinary precautions are taken, and many photos
and videos confirm it.This is in full contradiction with the assurances of the Annex D7 of the WHO joint-team
study report where she is instead on record saying that ‘all fieldwork is done with full PPE’.
1This distinct possibility was for instance mentioned by Lin Fa Wang, a collaborator of Shi Zheng Li on many
bat coronavirus research papers and a regular visitor at the WIV.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 4/20

As shown in the last row of the above table, the candidate population for an index case may go well
beyond the people directly working in a laboratory suite.
Actually, in the case of afield sampling infection (scenario LS1 of the table), it is the field sampling
personnel (strictly speaking any personnel at the sampling site) who need to be tested. Also because of
the high risk of transmission of SARS-CoV-2 via aerosols, not just the laboratory personnel and
students should be tested to detect a potential index case, but potentially anybody within the lab
institution or downwind from it, and anybody potentially exposed to lab wastes, be it solid or liquid,
inside or outside of the lab institution - as covered by the scenarios LS2 and LS3 of the table.
4. Which testing methods are relevant?
Active-infection (symptomatic) tests, such as PCR tests and IgM tests, are used to detect an
infection within afew weeks from the initial symptoms, but will not reliably return a positive test beyond
this initial period.
● PCR tests for SARS-like viruses (also called viral tests) determine the actual viral load present
in samples. They willl generally return a positive test up to 3or 4 weeks after initial symptoms,
and in some cases for longer. They are the most accurate test for active infection.
●IgM tests are antibody tests that can detect an infection 2 to 6 weeks (roughly) from initial
onsets.
Thus PCR or IgM tests would not be of any great use to detect an infection prior to the official first
outbreak date (8th Dec) if performed past mid January.
PCR tests and IgM tests would thus only be relevant to the detection of a possible infection before the
official start of the outbreak (8th Dec 19) if they were performed before approximately mid January 20.
In that case a positive test could relate to a possible lab-related accident prior to Dec 8th.
Fig 1: Levels of virus and antibodies after catching COVID-19
and the likelihood they will be detected during testing [source: health.govt.nz]
DRASTIC research - Steven Quay & Gilles Demaneuf p. 5/20

Post-infection (recovery) tests, such as IgG antibody tests, allow one to detect cases amongst
laboratory staff or students weeks or months after a possible infection. If these tests are done after the
official start of the outbreak (Dec 19) there is a strong possibility that such infection may not be due to a
lab-related accident but to a community-acquired infection during the later outbreak.
Standard IgG tests targeting SARS-Cov-like viruses are broad enough to be able to detect
SARS-CoV-2, even back in 2019. Some studies show that standard IgG tests are still relatively reliable
many months after infection, returning reliable positives 6 to 12 months after infection.
In conclusion, in order to detect infections in a Wuhan lab personnel or students, or in some field
sampling team, before the official breakout in Dec 2019, the tests of interests could be:
● SARS-CoV-Like IgM antibodies tests performed in 2019 or the first 2 weeks of 2020
(approximately).
● Post-infection SARS-CoV-Like IgG antibodies tests performed in late 2019 or anytime in 2020,
followed by necessary tracing to understand whether they were pre-breakout or post-breakout
infections.
Box 1: How long does a positive IgG test persist?
Afew studies demonstrate conclusively a sustained IgG positive outcome for over six months after the
start of the recovery phase.
● A paper in Lancet (Seroprevalence and humoral immune durability of anti-SARS-CoV-2 antibodies
in Wuhan, China: alongitudinal, population-level, cross-sectional study), based on patients in
Wuhan, concluded that:
Although titres of IgG decreased over time, the proportion of individuals who had IgG antibodies did
not decrease substantially (from 30 [100%] of 30 at baseline [April 14-15, 2020] to 26 [89·7%] of
29 at second follow-up [Oct9-Dec5] among confirmed cases, 65 [100%] of 65 at baseline to 58
[92·1%] of 63 at second follow-up among symptomatic individuals, and 437 [100%] of 437 at
baseline to 329 [90·9%] of 362 at second follow-up among asymptomatic individuals).
In other words 90% of individuals had sustained IgG antibodies 6 months at least after the first test
in April 20.
● A paper in Science (‘Immunological memory to SARS-CoV-2 assessed for up to 8 months after
infection’), based on COVI-19 patients in the US, reached very similar conclusions that:
‘The percentage of subjects sero-positive for RBD IgG at 6to 8 months PSO was 88% (35 out of
40) [--]The percentage of subjects seropositive for spike IgG at 6to 8 months PSO (≥178 days)
was 90% (36 out of 40).
The actual length of immune answer detectability was also found to depend on the individual.
● A British study found very similar results, with sustained Spike, RBD and nucleocapsid (N) IgG
positive rates for months.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 6/20

5. Which institutions should have been included in these tests?
BatCov studies were done routinely in P2 and P3 laboratories, not in the much more constraining P4
environment. For instance the National Virus Resource Center (NVRC, 国家病毒资源库), which is
affiliated to the WIV, used to list the relevant biosafety level for different pathogens before it was taken
down (with all the WIV databases it used to give access to), as per archived page.The list includes
under the heading ‘Animal Viruses’:
●蝙蝠冠状病毒 BSL-2 [Bat Coronavirus BSL-2]
●大鼠冠状病毒 BSL-2 [Rodent coronavirus BSL-2]
Practically this means that the China-WHO report is rather misleading as to the relevant BSL levels in
its evaluation of the possibility of a lab-related accident, never mentioning BSL-2 where some of the
BatCoV research took place, but mentioning instead BSL-4 where no such research was known to take
place:
“The three laboratories in Wuhan working with either CoVs diagnostics and/or CoVs isolation and
vaccine development all had high quality biosafety level (BSL3 or 4) [sic] facilities that were
well-managed,
(‘Argument Against’ p 119)”
As per the following study, these institutions are of particular interest for their demonstrated
involvements with BatCovs research at P2 or P3 level:
● WIV at its two sites: Xiaohongshan Park and Zhengdian.
● Wuhan [city] CDC
● Hubei [provincial] animal CDC
● Huazhong Agricultural University (HZAU)
● Wuhan University
● Wuhan Institute of Biological Products3
Fig 2: Biosafety laboratories in Wuhan [source: rdemaistre.medium.com]
3The WIBP is part of an integrated vaccine development platform which includes the WIV at Zhengdian. For
a detailed review of the WIBP activities, past and present, and the reasons for its inclusion see this study.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 7/20

6. Which laboratories are mentioned in the China-WHO report?
6.a Laboratories mentioned in report:
The annexes of the China-WHO joint team report mentions the following laboratories. In blue are the
ones which were asked about their testing of staff and students.
Annex D5
Annex D6
Annex D7
● Hubei [provincial] CDC
● Wuhan [city] CDC
● Jianghan [district] CDC
●Hubei [provincial] Animal
CDC4
● Wuhan Institute of Virology
Fig 3: Joint-team itinerary [source: WHO]
4As per the Annex D6 of the WHO joint-team study report the Hubel Animal CDC has a BSL-3 laboratory which is
not used due to the lack of staff (on top of its BSL-2 facility). Such lack of operational utility (including
maintenance) due to budget constraints is unfortunately rather common and has been a constant subject of
concern amongst Chinese experts. For instance Yuan Zhiming, director of the WIV P4, raised this precise issue as
recently as October 2019.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 8/20

6.b Comparison to target list:
Let’snow compare that list to the list of relevant labs and institutions as per 3. Above. In bold are the
common institutions.
Institutions of interest for a tests review
Institutions for which tests are mentioned in the
China-WHO joint-team report
●Wuhan Institute of Virology (two sites:
Xiaohongshan Park and Zhengdian)
● Wuhan CDC
● Hubei [provincial] Animal CDC
● Huazhong Agricultural University (HZAU)
● Wuhan University
● Wuhan Institute of Biological Products
● Wuhan Institute of Virology
● Hubei [provincial] Animal CDC
● Wuhan [city] CDC
● Jianghan [district] CDC
We can see that 3 of the 6 institutions of interest were asked about their tests.
6.c Misleading mention of Jianghan CDC tests:
We further note that the mention of the testing at the Jianghan [district] CDC in the WHO report can be
rather misleading as it makes strictly no sense. It is mentioned by Peter Embarek as an example of
testing that should dispel any idea that infected lab workers may have been responsible for the
outbreak.
However the tests conducted at the Jianghan CDC were viral [PCR] tests performed in June 2020 -
hence active infection tests which are of no relevance at all for detecting infections as far back as 2019.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 9/20

7. Who was actually tested?
7.a Limitations of the purported testing
The statements (see p. 1) made by the WIV or the joint-study about the testing procedure at various
Wuhan labs are rather vague and somewhat contradictory. Here is how these three statements
describe the populations tested:
● S1: staff and students in the [WIV] lab
● S2: staff from the WIV and three other institutions
● S3: workers and students from the laboratory of Zhi Zhengli at the WIV
In Statement S2, Peter Embarek explains that joint-study team asked the same question and got the
same answer from the four labs they visited while in Wuhan:
Putting all this back in our Lab-scenario table, we get:
Lab-Related
Accident
Scenario:
Field sampling accident
Lab acquired infection (LAI)
of Wuhan lab personnel
Lab escape without LAI
LS1
LS2
LS3
Entity of
Interest
Any field sampling team
Any institution in Wuhan working
on BatCoVs (may include WIBP)
All BSL levels
Any institution in Wuhan
working on BatCoVs (may
include WIBP)
All BSL levels
Entity tested
No mention
● WIV
● Hubei [provincial] CDC
● Wuhan [city] CDC
● Jianghan [district] CDC [but
active-infection tests
irrelevant]
No mention
Who should
be tested?
Anybody present during
field sampling
●Direct Infection in lab suites:
All people with access to the
lab suites and lab wastes
Indirect Infection via
contact: anybody sharing
space/facilities with people
working directly in lab suites or
handling the lab wastes inside
the institution.
●Indirect infection via
aerosols: Anybody within the
lab institution
●Indirect Infection via
contact: anybody in
proximity to any liquid wastes
from the institution or
handling the lab wastes
outside of the institution.
●Indirect infection via
aerosols: Anybody in
downwind proximity of the
institution.
Population
tested
No mention
No number given, just some
staff/workers and students.
Exact lab suites not clear
(beyond SZL lab)
No mention
Table 2: expected vs. actual entities and populations tested
DRASTIC research - Steven Quay & Gilles Demaneuf p. 10/20

In particular we note that:
● There is no mention of any testing done that would cover a field sampling accident (LS1). And
as per note (1) full PPEs were commonly not used, only ordinary protection (latex gloves, light
mask, and plastic poncho inside a cave)
● There is no mention of any testing that would address the possibility of aLab Escape with LAI
(LS3).
● At least two more institutions with a history of work on BatCoVs should have been considered:
the Wuhan University (A/BSL3) and the Wuhan Institute of Biological Products (WIBP).
7.b Target populations for the institutions mentioned
Let’s try to evaluate how many people should have been tested in any case in the labs mentioned in
the statements under the Lab acquired infection (LAI) scenario.
WIV Population
According to the official website of the Institute of Virology in August 2020, the Institute normally hosts
325 graduate students and 268 employees. That gives us atotal of around 593 people. As is known,
the Chinese government shutdown all “non-essential businesses” in Wuhan from February 13, 2020 to
April 8, 2020 (see Wuhan Lockdown Timeline). It has not been disclosed how much of the WIV was
closed during that time, but clearly those WIV staff who were not working during the lockdown but
instead were staying home should have shown the COVID-19 prevalence of the general Wuhan
population.
It is worth noting that the WIV has two sites:the old site (Xiahongshan in Wuchang district) and the
new site (Zhengdian in Jiangxia district) where the P4 is (with some P2s and P3s). However it is not
clear whether Shi Zhengli was referring to both sites when discussing sampling staff and students in
statement S1, or to the Zhengdian site only.
It is not clear also whether Shi Zhengli referred to all the labs that may have handled some BatCoV (be
they P2, P3 or even P4) or to her lab only. Given that BatCoVs were normally handled at P2 and P3,
and to avoid any suspicion on the P4, at the very minimum all the P2 and P3 laboratory handling
BatCoVs plus the P4 should be tested - this would cover a direct laboratory exposure.
But a direct lab exposure is not the only possible exposure. As SARS-CoV-2 is highly transmissible via
aerosols, and since it may also be transmitted via common surfaces and liquid wastes, a primary case
could happen out of one of the laboratory suites handling BatCoVs.This is something which happened
for instance during the Beijing Lab Leaks on 2004, when an ordinary electron microscope room (no
biosafety level) got infected by aonly partially inactivated SARS virus belonging to a P3 lab suite, with
the additional twist that the virus was actually held in a fridge in the corridor (no biosafety level again)
due to overcrowding of the lab suite.
To be able to understand how comprehensive the testing should be in the WIV,we would therefore
have to know how the laboratory suites are configured within the Xiahongshan or Zhengdian site,
whether these suites share common facilities with non-lab activities on each site, what the movement
of people is within each site and between the two sites, etc.
At worse it may thus mean that every single staff or student within both sites should logically have been
tested, whether working in alab suite or not, and irrespective of whether they were at the Xiahongshan
or Zhengdian site - since both sites were likely working on coronaviruses in the BSL-2 or BSL-3 labs.
That would mean 590+ staff and students.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 11/20

Strictly speaking we may have to also consider anybody with a prolonged physical presence in these
institution precincts - given the potential risk of infection via aerosols. That would include any temporary
construction workers on site, such as the ones using the temporary barracks that were only within 15
meters of the P2/P3 labs in the WIV compound (see page 30 of this report).
Four Wuhan labs (as per S4)
We do not have enough information to venture how large a candidate population for testing the four
labs mentioned in statement S4 may represent. If we suppose very roughly maybe 50 people
(researchers, staff and students) potentially exposed to a direct laboratory infection in each one of
these 4 institutions, and another 50 in each institution potentially exposed to indirect infection via
aerosols and shared facilities or wastes, we get already 400 persons in total that should be tested as
priority over these 4 institutions.
Note that the exact Wuhan institutions and laboratories involved in coronavirus research - by opposition
to the ones mentioned in statement S4 - were listed in section 4.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 12/20

8. Observed seropositivity rates in WIV population of interest in April 2020
In order to estimate how many employees and students at the WIV should have returned a positive
antibodies test in March 2020 (as per the Science Magazine statements), we shall first make use of this
paper (Oct 20) which establishes background IgG antibody seropositivity levels in April-May 2020 in
Wuhan residents through a cross-sectional study of around 35,000 individuals.
Note that that paper only considered individuals without a COVID-19 history - leading actually to a
slight underestimation of the seropositivity rate across the full population, which additionally includes
people with a COVID-19 history - hence people with a likely much higher IgG positive rate.
In a very useful manner, the paper gives seropositivity prevalence levels for IgG testing by urban/non-
urban districts, age and sex. It further gives a confidence interval (C.I.) for the prevalence level in the
urban areas.
We thus obtain a4.4% IgG+ prevalence level for urban Wuhan in April 2020, with a95% C.I. of [4.0%,
4.8%].
‘The seropositive prevalence in the urban districts was higher than that in the suburban and rural
areas (4.4% [95% C.I.: 4.0%-4.8%] vs 2.9% [95% C.I.: 2.3%-3.6%]; P < .001), demonstrating an
urban to suburban gradient’
Fig 4.a: Severe Acute Respiratory Syndrome Coronavirus 2 Seropositive Prevalence by Sex and Age
Group [source: jamanetwork.com]
Fig 4.b: SARS-CoV-2 Seropositive Prevalence in Different Age Groups
[source: jamanetwork.com]
As a first pass, as most representative of the WIV population of interest we shall use the urban district
IgG positive (IgG+) rate of 4.4% in April 2020.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 13/20

9. Expected seropositivity rates in WIV population of interest in March 2020
We now need to assess what the same prevalent IgG+ rate for Wuhan urban would have been just one
month earlier, in March 2020.
To do so we observe that there were very few COVID-19 cases with onset date past the 20th Feb 2020
and that IgG+ rates start flattening out at week 3 past onset. Hence the people who would have tested
IgG+ in April cases would have in their large majority (90%+) also tested positive in March - with only
some of the cases with onset in the last 3 weeks of February potentially not testing IgG+ in early
March.
Fig 5: Daily numbers of laboratory-confirmed and clinically-diagnosed cases
[source: joi200040supp1_prod.pdf ]
Conversely, given that various studies (see box 1) show that the vast majority of IgG positive
individuals (90%+) would again test positive up to 9 months (if not more) after infection, then the vast
majority of the people who would test positive in March would again test positive in April.
Hence the individuals who would test positive in April would also test positive in March (except for a
small minority of late February onset cases), and vice-versa. We shall thus conservatively round down
the 4.4% April IgG+ prevalence rate to a 4% March IgG+ prevalence rate.
We shall later consider a few alternatives in a sensitivity analysis.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 14/20

10. How many of the WIV employees and students should have tested
positive in March 2020?
Based on the above background seropositivity level we shall calculate the expected number of returned
seropositivity tests at the WIV if they had been done in the period covered by the study: March 20.
To do so we will use a binomial distribution. That probability model assumes observations (draws) are
independent of each other with the occurrence of a positive result having the same probability, p. The
important point here is 'independent'. If there was an infection cluster in alab, then that assumption
would not be appropriate.
However, that is perfectly fine. We shall suppose no infection clusters in the lab and calculate how
many employees/students should be expected to be seropositive for IgG just by virtue of the Wuhan
background seropositivity level applicable to that population of interest, assuming a similar level of
community-acquired infection for WIV staff. Any cluster at the WIV would in any case increase that
rate, so at worse we will underestimate the true rate this way,
The probability of having x seropositive results in n individuals, supposing independence of the draws
(with p the probability of a positive result and q = 1-p the probability of a negative result), is given by the
binomial distribution:
and the mean is simply np.
10.a Base case:
Using 593 employees and students (n) at the WIV (across both sites) and 4% Wuhan community
prevalence rate in April-May 2020 (p) will find that:
● We should expect around 24 positive tests at the WIV (0.04 x 593),
●The probability of returning no IgG positive test at all (x=0) at the WIV in March 2020 is ~
0.000000000003 (11 zeros after the comma), basically null.
Fig 6.a: Binomial Distribution (x=0, n=593, p=4%) [source]
DRASTIC research - Steven Quay & Gilles Demaneuf p. 15/20

Even assuming testing of only 51 employees and students, effectively a very small subset of the
population that could include the index case for a lab-related accident, the probability of returning no
IgG positive test at all is only 12.5%, and is halved with every additional 17 individuals.
Fig 6.b: Binomial Distribution (x=0, n=51, p=4%) [source]
10.b Sensitivity analysis:
To check how robust this result is, we shall consider a few variations based on the applicable
background seropositivity rates. The 4.0% we noted above should be quite stable, being calculated on
a fairly large population -but let’s assume that it may be as low as 1.00% for illustration [an extremely
low and very unlikely value, well beyond the 95% C.I. of roughly 3.6%-4.4%].
Even with 1% background seropositivity rate the conclusion remains essentially the same: the
probability of returning 0 positive tests is negligible at 0.26%.
probability of 0 positive test
average number of positive tests
bgr inf. \ WIV pop
593
bgr inf. \ WIV
pop
593
1.00%
0.0026
1.00%
6
2.00%
6.27E-06
2.00%
12
3.00%
1.43E-08
3.00%
18
4.00%
3.07E-11
4.00%
24
4.40%
2.58E-12
4.40%
26
Conclusion:
Had the WIV population been tested in May 2020 it would have without any doubt returned positive
cases. On average 24 with an assumed positivity rate of 4%.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 16/20

Box 2: Mathematical corner
Here is how you can easily tabulate the probabilities in your head.
The key is to remember that with a4.0% seropositivity rate there is very close to a50% chance of no
seropositive in a group of 17 people. This is a simple binomial distribution result. See for instance this
online calculator.
If you add another 17 people, the probability of no seropositive in the group is again divided by 2. From
there you can see that the probability of no seropositive at in a group of (n * 17) people is ~1/2^n.
510 is still short of the 590 staff and students at the WIV,but conveniently 510 =30 x17 = 3 x 10 x17.
Also 1/2^10 ~1/1,000, so the probability of no IgG seropositive in a group of 510 people is around
1/1,000^3 = 1 in a billion!
If one prefers to work in powers of 1/100 then the probability is about 0.01 for a group of 113 people.
So for 565 people (5 x 113) it becomes 1/10^10, a 10th of a billionth.
DRASTIC research - Steven Quay & Gilles Demaneuf p. 17/20

11. How many of the WIV staff & students did get infected by March 2020?
Let’snow move from seropositivity to infection by evaluating the expected infection rates in the WIV
population of interest in March 2020:
If we ignore false positive and false negatives, the seropositivity prevalence levels above tell us that at
least 4% of the WIV population of interest should have been infected with SARS-CoV-2 by May 2020.
The reality is that there should be some false positives and false negatives. False negatives cause an
underestimation of the actual prevalence level, while false positives cause an overestimation. To be
conservative we shall only consider false positives.
The relevant confusion matrix of the IgG test is thankfully given in the supplement of the study:
“The sensitivity, specificity, positive predictive value, and negative predictive value of IgG were
87.2 %, 99.3%, 98.8% and 91.7%, respectively.”
Let’s include the false positives in the analysis, while ignoring the false negatives - a very
conservative assumption. Applying the Positive Predicted Value (98.8%) to 4% of seropositives
(and), we get a rate of True Positive of 3.95%.
Stat
Value
PPV
98.80%
TP/(TP+FP)
NPV
91.70%
TN/(TN+FN)
Sensitivity
87.20%
TP/PC =TPR
Specificity
99.30%
TN/NC = TNR
We could go further and include the false negatives to get the exact True Positive rate. This is done in
the Supplement to this paper. However because the resulting rate is rather high (possibly due to issues
with the confusion matrix of the IgG test), we prefer to conservatively use the 3.95% above.
From there we can deduce a conservative estimate of the number of WIV staff & students likely
infected by March 2020.
Based on the conservative estimate of 3.95% for the True Positive Rate, the expected number of staff
and students likely infected with COVID-19 by March 2020 is essentially the same as the expected
number of positive tests (24).
DRASTIC research - Steven Quay & Gilles Demaneuf p. 18/20

12. Some possible interpretation
To be very clear, the fact that some personnel and students should have tested positive to BatCoV
antibodies, at the WIV and other key labs around March 2020, does not necessarily mean that the
pandemic began with a laboratory-acquired infection at some time in 2019. But such an ineluctable
statistical conclusion of necessary infections, if it had been confirmed by the WIV and other labs, would
have led to proper tracking and review of these cases - the absolute minimum requirement when
investigating an outbreak.
What is surprising is that for some reason Chinese officials seem to have preferred avoiding any such
standard line of epidemiological investigation by making statements that do not bear scrutiny today.
Here we will venture a possible interpretation based on the available information at the time.
When Dr. Shi was asked in July 2020 (statement S1)about whether any staff at the WIV tested positive
for COVID, the data available at the time was essentially that only around 50,000 COVID-19 cases had
been recorded in Wuhan, with yet no broader seropositivity study being published.
Precisely 49,912 official COVID-19 cases over a Wuhan population of 11 mln (as of end March 2020,
and little changed after that) could be interpreted as a very low prevalence rate of around 0.454%. Only
proper IgG testing across large urban cohorts would later convincingly show that the real prevalence
rate within Wuhan was actually much higher; 10 times higher at about 4.4%, a rate which is still low but
much closer to the rates recorded in many cities worldwide (for instance Madrid: 11.3%, Boston:9.9%,
Stockholm: 7.35%, all in May 2020).
Applying the binomial distribution to this ‘prevalence’ rate of 0.45%, we can see that the likelihood of no
test returning positive over 51 staff members is around 79%, which is a much more credible rate than
the 12.5% likelihood arrived at using the actual background rate of 4% .
It is therefore conceivable that Dr. Shi was only aware of this low rate of around 0.45% when she made
that statement in July 2020. A combination of a (i) low number of staff/students tested (focussing -
wrongly - strictly on her team) and (ii) such an underestimated prevalence rate of 0.45% would indeed
give a credible probability (of 79%) of not returning any positive IgG test.
For the sake of clarity, this does not mean that Dr Shi was purposely lying. The exact information about
the results of the various tests would have been carefully controlled, and it is very possible that Dr Shi
was simply repeating what she was told and believed at the time, even if a simple analysis - as done
here - should now convince her of the improbability of these results.
Fig 11.a: Binomial Distribution (x=0, n=51, p=0.454%) [source]
DRASTIC research - Steven Quay & Gilles Demaneuf p. 19/20

Fig 11.b: Binomial Distribution (x=0, n=593, p=0.454%) [source]
DRASTIC research - Steven Quay & Gilles Demaneuf p. 20/20