Coronavirus Fact-Checking Taskforce
This is an open-source summary of the scientific information about common COVID-19 questions. If you want to participate, please see the last section “How to participate”. Check our Facebook group for updates and to ask questions! Visit also our partner: Adios Corona.
Warning: Many articles cited here are not peer-reviewed (published as preprints). Given the emergency, we reviewed them for obvious mistakes. Keep in mind that such data is of lesser quality, and it can change or be retracted.
Disclaimer: the information here is gathered by a team of volunteers. It is not necessarily up to date and it is your responsibility to use that information with caution, read the sources and decide for yourself. No matter what it looks like, we do not make recommendations and are no official entity. We simply provide curated information that is susceptible to error.
In this table, we compare the properties of the three big families of masks: N95-type (FFP2, KN95...), surgical/medical masks and cloth/home-made masks. Note that the material and number of layers for the home-made mask is critically important: 1-layer silk, cotton 80TPI or gauze had extremely low filtrations .
A good mask would filter particles of the size of aerosol droplets (nuclei) that are <5μm (typically measured by bacterial filtration efficiency BFE and PM2.5) and particles of the size of the virus (after droplet evaporation), which is 100 nm (0.1 μm) in diameter. The mask also needs to be breathable and to have a good adjustment to the face (fit factor). Finally, the mask should have demonstrated a reduction in the number of particles shedded during coughing or breathing, or in the risk of catching flu-like or SARS-like diseases.
Filtration efficiency with no leaks
~0.1 μm particles
1 layer 600TPI cotton: 70% 
1 layer 600TPI cotton + 2 layers silk: 95% 
1 layer tea towel: 72% 
1-3 μm particles
1 layer 600TPI cotton: 98% 
1 layer 600TPI cotton + 2 layers silk: 97% 
1 layer tea towel: 83% 
Filtration with leaks
1 layer 600TPI cotton + 2 layers silk: 35% 
3 μm particles
1 layer 600TPI cotton + 2 layers silk: 35% 
Breathability (pressure drop in Pa - lower is better)
1 layer tea towel: 7.2 
90  (adults)
20  (children)
5  (adults)
4  (children)
2  (adults)
2  (children)
Number of air particles shedded with mask
Receiver: -57–86% 
Emitter: -95.5% 
2-layer cotton: -83% 
Receiver: -17–37% 
Emitter -57% to -76% 
Effect on SARS infection risk relative to no mask
-45% to -60% 
-62% to 75% 
-45% to -60% 
 Rengasamy et al., 2017 (NIOSH NaCl method - no leak - 0.1 μm average)  Rengasamy et al., 2017 (FDA-PFE method - Polystyrene latex sphere test - 0.1 μm - no leak)  Rengasamy et al., 2017 (ASTM-BFE method - Bacterial filtration efficiency with Staphylococcus aureus - 3 μm - no leak)  Konda et al., 2020 (NaCl method - size range 0.01-10 μm - with or without leaks - flow rate 33L/min)  Offeddu et al., 2017  Chu et al., 2020 (not COVID but SARS, ILI and influenza)  Davies et al., 2013 (aerosol of Bacteriophage MS2 virus - 0.02 μm - and of Bacillus atrophaeus - 1 μm - no leak - not clear what unit is the pressure drop - shedding after growth for particles <5μm)  Anderegg et al., 2020 (NaCl method 0.2 μm - no leaks)  Oberg & Brosseau, 2008 (latex sphere test 0.9 μm and 3.1 μm at 6L/min - no leak)  Oberg & Brosseau, 2008 (NaCl method 0.07 μm at 84L/min- no leak)  Chan et al., 2020 (hamsters using sars-cov-2 - no leak)  Viscusi et al., 2009 (NaCL method 0.3 μm at 83L/min - no leak)  Song et al. 2020 (PM2.5 method - 2.5 μm - no leak)  Fisher et al. 2020 (laser beam method - >0.15 μm - with leaks)  Ueki et al. 2020 (sars-cov-2 shedding after growth - with leaks - at 50cm)  Milton et al. 2013 (particles <5 μm after growth)  Mueller et al. 2020 (NaCl method - with leaks - particles <0.3 μm)  van der Sande et al. 2008
It is possible to drastically improve the fit of a surgical mask using the double-eight method (using three rubber bands and one paperclip): https://onlinelibrary.wiley.com/doi/full/10.1002/emp2.12335
For a more detailed review of the literature on the filtration aspect, comparing surgical mask and cloth mask: https://www.mayoclinicproceedings.org/article/S0025-6196(20)30826-0/fulltext
Certainly Yes. In a recent meta-analysis, wearing eye protection (face-shield or goggles) decreased the risk of catching acute respiratory syndromes (16%) by 65% (to 5.5%) . After a cough, ~0.85% of the produced particles are inhaled by someone located at 50 cm wearing no mask . Immediately after a cough at 50 cm, it stops 96% of the viral charge and reduces the virus on a mask by 76-97%, which drastically eases the mask decontamination for reuse . The face-shield can also potentially decrease the contamination by the eyes, which can be used very well by that virus , and avoid touching the face by mistake. However, it does not replace a mask because only 30% of the airborne particles are stopped .
Likely yes. Surgical masks are the second-best masks. They able to filtrate 75% of 0.1-μm particles  (Sars-Cov-2 has a size of 0.1 μm ), or 55-88% (different models) to 98% when using different methods , or 90% of a virus which is 4 times smaller than the Sars-Cov-2, in a device with no leak . With typical leaks due to face fit, filtration dropped to 50% . For larger particles (1-10 μm), filtration was 75% with (but see  for an outlier result of 0% filtration). An observational study of elementary school children in Japan found that children wearing masks decrease by 16% the risk of catching seasonal influenza . In recent meta-analyses, wearing a surgical mask reduced the risk of SARS by ~80%  and various acute respiratory syndromes by 45-60% . A randomized study found that people were 70% less likely to catch a respiratory infection when they were wearing a mask . In addition, they can be useful to avoid spreading the virus (see below). Filtration with surgical masks were superior by 15 points compared to typical cloth masks (made of 2-layer cotton t-shirts, tea towels or cotton 600 TPI). Whether surgical masks can replace a N95 respirator is possible but unclear: in two meta-analyses, healthcare workers were as ill when using a N95-respirator as when using a surgical mask (Bartoszko et al., 2020, Long et al., 2020). Another review contradicted the meta-analysis but had clear conflict of interests (MacIntyre & Chughtai, 2015).
 Offeddu et al., 2017  Chu et al., 2020  Rengasamy et al., 2017 (used 4 different methods - no leak)  Davies et al., 2013 (Bacteriophage MS2 virus - 0.025-μm particles - no leak)  Quan et al., 2017 (2.5-4 μm-diameter aerosols but H1N1 is 0.1μm - only the middle layer)  Bar-On et al. (2020)  Konda et al., 2020 (NaCl method - various size 0.01-10 μm - with or without leaks)  Uchida et al., 2017  MacIntyre et al., 2009
Important database of scientific reviews: https://www.n95decon.org/
Reviews of surgical masks not working for healthcare workers (biased, read carefully): https://www.rcreader.com/commentary/masks-dont-work-covid-a-review-of-science-relevant-to-covide-19-social-policy
But see: https://www.phc.ox.ac.uk/publications/1102534, https://www.journalofinfection.com/article/S0163-4453(20)30235-8/pdf and https://rs-delve.github.io/reports/2020/05/04/face-masks-for-the-general-public.html
An experiment showing that people are failing to wear their mask correctly to be efficient: https://www.sciencedirect.com/science/article/pii/S0196655307007742
Likely Yes. Surgical masks can reduce detectable viral shedding in respiratory droplets and aerosols of symptomatic individuals with coronavirus (Leung et al., 2020) and are useful to avoid dissemination from the wearer, decreasing by 6 the number of cough-produced particles and microorganisms (Davies et al., 2013).
Further reading: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3591312/
A recent meta-analysis: https://www.thelancet.com/action/showPdf?pii=S0140-6736%2820%2931142-9
Certainly Yes. Surgical masks filtered 85% of 2.5-4μm-diameter aerosols containing the H1N1 virus after coating it with a salted solution (Quan et al., 2017). What is true for H1N1 is likely for Covid-19. It also resulted in a 100%-survival rate for mice wearing the filter (against a 0%-survival rate when mice wore non-coated masks). The coating simply consists in leaving the mask in the salt solution (~27g of salt for 10 cl of water) with a drop of surfactant (liquid soap) for a night and then letting it dry on the heater (not under the sun).
Certainly Yes. N95 masks were impregnated with 2.0-2.2% (w/w) of copper oxide in 3 out of 4 layers (Borkow et al, 2010). Their filtration level was similar to the control N95 masks (99.9%-99.99%) but the non-filtered viral contagiousness of human influenza A virus (H1N1) and avian influenza virus (H9N2) was reduced by ~99.999%. These masks filter bacteries (%BFE) and show breathability (ΔP) as good as the EN14683:2005-type-IIR and NIOSH-N95 standards. What is true for the used viruses is likely for Covid-19. The authors also reported that the amount of copper eluted into the air downstream of the tested masks after 5 hours was within the safe exposure limit, and therefore deemed these masks safe for human use. The study did not investigate the reusability of these masks.
Likely yes. See full documented answer at Adios Corona here.
A safe solution is leaving the mask in an envelope or paper bag with the date on it for 7 days because only 0.01%-0.1% of the virus can survive 7 days on the mask . There is minimal risk of bacteria/fungus growth given that the paper will dry the mask and bacteria/fungus need water to develop. Leaving it on the heating will be even safer given the virus is sensitive to heat . The recommended duration of a surgical mask varies by countries. In Hong Kong and China, no more than 4 hours is recommended before disposal by experts (Ho, MingPao, 7 January 2020; Chung, HK01, 7 February 2020). In Taiwan it references the guidance in the US, and recommends a duration of no more than 8 hours cumulatively (National Taiwan University Public Health College, SARS Promotion Material, Principles for using Masks). However, this recommendation is only based on a WHO report  citing a study  showing that mask tolerability (not efficacy) is decreased after several hours (see more on Adios Corona). Indeed, when the filtration is tested several times after mimicking the use with an alternating air flow, it is little affected by prolonged reuse or multiple 4-hour cycles, with an average loss of 0.3% by cycle . It was only tested for 3 cycles on surgical masks but given the filtration loss is identical to a cycle with an N95, it is likely that can extrapolate from the results on N95 : after 30 cycles, an average of 9% filtration loss brings the surgical mask efficiency close to the cloth mask. Fit factors were not affected by usage .
Recent studies indicated that ‘dry-steaming’ may allow these masks to be reused one to five times (see Sterilization section) or heating it for 1h at 70°C, which seems safe although limited to limited evidence exist for reusability (filtration, fit factor) under those conditions (see Sterilization section). Heating it for 30min at 70°C significantly decreased filtration efficiency after 3 cycles but the filtration was still above 90% . Bleach will certainly leave smell and possibly toxic waste (Viscusi et al., 2009). UV sterilization might be inefficient because UV light cannot reach the middle (filter) mask layer. Salt or copper coating has been proposed but without evidence that multiple cycles does not degrade mask filtration or resistance (Quan et al., 2017).
Certainly Yes. (adapted from ). N95 / FFP2 masks are the best masks. They are designed to filter 95% of the 0.3-μm particles. In fact, they are filtering 90%  to 99.8% of 0.1-μm particles  (which is Sars-Cov-2’s size) and 99.9% of 3-5 μm particles . Therefore, N95 masks are capable of filtering most free virions, but also droplets produced by sneezing and droplet nuclei (5 μm aerosols), and they provide filtration efficiencies larger by 15 points than surgical masks, along with a better face fit.
In recent meta-analyses, wearing N95 respirators reduced the risk of SARS by ~80%  and various acute respiratory syndromes by 90-95% . N95 protects better than cloth masks with 13 times more influenza among hospital staff using cloth masks (2-layer cotton) than among those using FFP2 masks . N95 masks are comparable to other international mask norms (see glossary). Note that respirators with a valve should be avoided if you are infected because the virus can spread through the valve when coughing.
 Offeddu et al., 2017  Chu et al., 2020  Rengasamy et al., 2017 (no leaks, various methods and particle sizes)  Bar-On et al., 2020  Konda et al., 2020 (NaCl method - various size 0.01-10 μm - with or without leaks)  MacIntyre et al., 2015
Certainly Yes. Reusable respirators (for FFP2, it is written FFP2 R on them) can be used continuously for 8h and reused at least 3 times without loss of filtration . Long-term use of N95 shows a linear average loss of 0.3% by cycle . Reusing a N95 affects very little the fit to the face (no change after 5 cycles ) but even after 30 cycles, the median fit factor is still at 100  (which is a good score - meaning a loss of filtration <1% ). Less than 0.1% of the virus can survive 7 days on the mask . Therefore, leaving it in a paper bag for 7 days (preferably on a heater) should be enough. For full sterilization, dry heat the respirator in the oven for an hour at 70°C. It is efficient and respirators retain filtration efficiency, breathability and mask fit for 10  to 20 cycles (Gertsman et al., 2020, Liao et al., 2020) although one study found reduced fit factors after only 3 cycles . Verify that the mask is not damaged after each cycle. All other sterilization methods should be avoided. Leaving the mask under the sun or UV light can damage mask resistance (Lindsley et al., 2015) and filtration (Viscusi et al., 2007). Alcohols (like isopropanol), bleach and moist heating substantially degrades filtration efficiency (Martin & Moyer, 2000 - cited from Lindsley et al., 2015, Viscusi et al., 2007). Two-min microwaving or an hour of dry-heat above 90°C melted the respirators (Viscusi et al., 2009). For more information, see the Sterilization section. You can also see how decontamination affects filtration efficiency, fit factor, and strap performance for various methods
 Tsai, 2020 (table 6)  Fisher et al., 2020  Kasloff et al., 2020  Moyer & Bergman (2000)  Daeschler et al., 2020  Bergman et al 2012  CDC decontamination results, oct 2020  Fisher et al., 2020  Anderegg et al., 2020  OSHA Respiratory Protection Standard 1910.134. (1998)
Certainly yes. Several N95 models retained filtration efficiency above 95% and breathability even 13 years after manufacturing (Lin et al., 2020, Viscusi et al., 2009). However, filtration efficiency is clearly decreasing with time, and the aging of the straps and seal materials of the N95 respirators may affect their fit factor. Indeed, 98% of 3971 expired respirators retained their filtration efficiency and breathability, in an NIOSH’s case report (Greenawald et al., 2020).
Summary: Home-made masks had various virus filtration efficiency depending on the material, all above 50% and lower than the surgical mask but were useful to decrease virus spread, although less efficient than the surgical mask to protect the wearer. A good compromise between efficiency and breathability could be the pillowcase material. Pocket tissue and kitchen roll tissue could also be useful against bacteria and possibly against viruses. For better results, one can combine layers, like a good filtration layer (high thread count cotton fabric or dish towel) and an electrostatic layer (natural silk, chiffon weave or flannel). You can add the additional filter like this.
Protection against respiratory diseases: In a recent meta-analysis, wearing a cotton-mask or a surgical mask reduced the risk of various acute respiratory syndromes by 45-60% .
 Chu et al., 2020
Filtration efficiency: Home-made masks of various materials all filtered more than 50% of aerosolized virus (the Bacteriophage MS2, which is 4 times smaller than the covid-19) when no leak could occur (Davies et al., 2013). The best was the mask made with a vacuum cleaner bag (86%-efficiency, close to the 90%-efficiency of the surgical mask) and the worst were the scarf and cotton t-shirt (49%-51%). Data is illustrated in the image below. Home-made cotton t-shirt masks had 3 times more leaks than the surgical mask (using the fit factor) (Davies et al., 2013). In a cluster-randomised controlled trial, healthcare workers using cotton masks were more ill than those using surgical masks (MacIntyre et al., 2015). Home-made masks were useful to avoid dissemination from the wearer, significantly decreasing the number of cough-produced particles (by 5, almost as good as the surgical mask) and microorganisms. The efficiency of home-made cotton masks could be improved by adding a layer of nylon stocking (Mueller et al., 2020, preprint). It could also be improved by making it a tighter fit around the face and by including a filter layer (Kwong, 2020; personal communication). A cotton mask sandwiching one to two sheets of pocket tissue or kitchen roll tissue could achieve a bacterial filtration efficiency (BFE) ranging from 45% to 91%, compared to the 98% BFE of an EN:14683-Type II surgical mask. Note that efficiently filtering a bacteria (high BFE) does not necessarily translate into efficiently filtering a virus. Finally, cotton fabrics with tight weavers and low porosity (e.g. high thread count, 600 TPI vs 80 TPI) can improve filtration efficiency (Konda et al., 2020). When there is no leakage, combining high thread count cotton fabric with one that could provide good electrostatic interactions, such as natural silk, chiffon weave or flannel, provides a filtration efficiency comparable to a surgical mask for particle sizes of <300nm (>80%) and >300nm (>90%) (Konda et al., 2020).
Breathability: Surgical masks are designed to have good breathability. The vacuum cleaner bag had the worst breathability when measured by pressure drop (Davies et al., 2013). Data is illustrated in the image below: as a trend, the better the filtration efficiency, the worst the breathability. However, no statistics were done to compare it to the surgical mask. Using differential pressure (∆P) as the indicator of comfort and breathability, cottons (1 or 2 layers, low or high thread counts), chiffon (1 or 2 layers), natural silk (1, 2 or 4 layers), and a hybrid of cotton and chiffon / natural silk / flannel, showed a value of 2.5 ± 0.4 Pa which indicates good breathability (Konda et al., 2020).
Additional information on breathability of HK Mask filter materials: https://drive.google.com/drive/u/1/folders/1-6P50te3sHHIutk85Qx4wgUAJ3Mjb6E6
A list of home-made mask designs:
A recent meta-analysis on the efficiency of masks: https://www.thelancet.com/action/showPdf?pii=S0140-6736%2820%2931142-9
Likely Yes. Among the tested materials in this study (Davies et al., 2013), all materials, beside tea towels, did not provide significantly better bacterial filtration efficiency (BFE) by doubling the fabrics. In contrast, a recent study showed that doubling up low thread count cotton (80 TPI), high thread count cotton (600 TPI), chiffon and natural silk could marginally improve the filtration efficiency for particles sizes of <300nm and >300nm (Konda et al., 2020). As for filter materials, using one, two and three layers of pocket tissue as the filter in a cotton mask case increased the BFE from 46%, to 70% and 83% respectively (Kwong, 2020; personal communication). Using two layers of kitchen roll tissues may also provide better bacterial filtration efficiency compared to just one layer. Note that efficiently filtering a bacteria does not necessarily translate into efficiently filtering a virus.
Maybe Yes. Based on expert opinion and tests performed (Kwong, 2020; personal communication), using a cotton mask case sandwiching a filter as an example, putting 2 layers of pocket tissue or kitchen roll tissue as filter with an orientation of 0 degree, 45 degree or 90 degree from each other improved the absolute BFE value by no more than 2%. Please note that efficiently filtering a bacteria does not necessarily translate into efficiently filtering a virus.
Certainly Yes. Given 99.999% of the virus on cloth is inactivated at 22°C after only 30 min (Chin et al., 2020), it is enough to simply wash the mask in laundry (if cotton, not a plastic-derived material) at 60°C (Kwong, 2020; personal communication). For extra-precaution, one can sterilise the mask by ironing. This type of home-made masks is recommended to be used for up to 4 hours without being taken off. If any, the filter (for example, made of vacuum cleaner filter, coffee filter or tissue paper) should be discarded after each use. For other types of home-made masks such as those on this list (Nugent, KNOWM, 1 April 2020), reusability varies and depends on each design. Please note that this is based on expert knowledge.
Certainly yes. See all the data on that entry of Adios Corona.
As a summary, wearing a mask decreases the inoculum dose  which is shown to decrease the severity of the disease, and the probability of symptoms, once contaminated .
Note for reviewers: For each sterilization article, report which conditions are used for sterilization (e.g. 1 min of microwaving at 1200W is very different from 20 sec at 400W) and whether 99.9% of virus is inactivated or more. For each mask reuse test, report which conditions are used, and all outcomes when possible. Indeed, we consider that a mask's sterilization method is useful if it does sterilize it, if it does not damage the mask (filtration efficiency, fit factor, or resistance) and if it does not leave toxic waste on the mask (Wiwanitkit, 2011). Note that N95, surgical masks and cloth masks should be considered as different entities.
Time to reduce concentration by
30 min 2
<3 hours 2
30 min2 - 1h5
3h2- 2 days5
3h2 - 6 h5
6h2 - 4 days5
30 min2 ,<1h6- 2h5
4 days 2
1 days 1
2h4, 8h8,9, 1 day2,5
14, 2days2,8,9 -55
24, 39, 5 days2,8 75
8h8,9, 1day1,3, 22,5,6
21,8,9,3days3,6, 72,5 days
Plastic (not polystyrene)
Inner layer of a surgical mask
7 days 2,5
Outer layer of a surgical mask
7 days 2
8 h3 -26 days
Model: 26 h3 - 216 days
In bold is the median of studies. 1 van Doremalen et al., 2020 (50 μl at 40% humidity, PP plastic), 2 Chin et al., 2020 (5 μl at 65% humidity), 3 Fisher et al., 2020 (50 μl at 40% humidity), 4 Pastorino et al., 2020 without BSA (50 μl at 50% humidity), 5 Liu et al., 2020 (50 μL at 35% humidity), 6 Kasloff et al., 2020 (with BSA, 10 μl humidity at 37%), 7 Pastorino et al., 2020 with BSA (50 μl at 50% humidity), 8 Hirose et al., 2020 (5 μl at 50% humidity and 25°C, with Eagle’s medium and fetal bovine serum) 9 Hirose et al., 2020 (same with human mucus). We do not show Riddell et al., 2020 in the table because this study is an outlier but we discuss it below.
Survival times (minimal times after which the virus is undetectable using TCID50) were tested under standard conditions (room temperature 22°C / 65% humidity), and after aerosolization for the air condition. Be aware that if the COVID-19 is as infectious as the common flu, a TCID50/mL of ~0.1 is sufficient to infect half of the people exposed (Couch et al. 1966).
Survival times could be longer when using BSA (Bovine Serum Albumin) mimicking bodily fluid according to Pastorino et al., 2020 (99%-decrease after 4 days on glass and polystyrene, after 3 days on aluminum) or Riddell et al., 2020 (10 μl at 50% humidity), but Kasloff et al., 2020 and Hirose et al., 2020 did not confirm these results.
For the airborne condition, note that nebulized virus is not a common state for the virus, so that although the virus was only reduced by 84% after 3 hours, one cannot conclude that the virus can survive long in the air in normal conditions (Peters et al., 2020).
Further reading: Limited contagiousness of virus on surfaces in patient households: https://www.medrxiv.org/content/10.1101/2020.05.28.20114041v1
More measures including skin: https://www.medrxiv.org/content/10.1101/2020.07.01.20144253v1
Possibility that dryness explains the survival on different surfaces: https://www.preprints.org/manuscript/202008.0426/v1
Possibility that mosquito-repellent can kill the virus: https://www.gov.uk/government/publications/experimental-survival-of-sars-cov-2-on-an-insect-repellent-treated-surface--3?fbclid=IwAR2I04CwmMDBpPiP4siel_OPxVKs7iopNEXpcpBJTRZRZIyZDwqh035js9E
Note: do not use only the data below to decide the technique to sterilize masks: their efficiency could be damaged by the sterilization process. See the relevant sections (N95/FPP2 respirators, surgical mask, home-made mask) to know which technique to use.
Keep in mind that it is likely that extremely small amounts of virus left can still be infectious. The level of reduction you should aim for (99.9% or 99.999%) crucially depends on your risk of exposition to the virus.
At 4°C, the virus is stable for at least 14 days, with no sign of decline, and in solutions of pH between 3 and 10 (Chin et al., 2020). All studies below measured TCID50/mL.
Time to reduce concentration by
Depends on the material - see survival on surfaces
Dry heat 22°C
Dry heat 37°C
Dry heat 56°C
Dry heat 70°C
Dry heat 70°C on a mask/steel
47 min (N95)2, 88 min (steel)2
Model: 1.15h (N95)2, 2h15min (steel)2
30%-ethanol (30° alcohol, e.g. vodka)
70%-ethanol (70° alcohol)
6 min (mask)2
30 sec3, <5 min1, 13 min (steel)2
Handrub WHO formulation I, >40%*
Handrub WHO formulation II, >30%**
2% soap solution
7 min (steel)2, 1 hour (mask)2
15 min (steel)2, 2 hours (mask)2
Vaporized hydrogen peroxide
7 min (steel)2, 10 min (mask)2
15 min (steel)2, 20 min (mask)2
Benzalkonium chloride 0.1%
3Kratzel et al., 2020, who used suspensions of virus with 1 part of organic load for 8 parts of the sterilization substance.
4Batéjat et al., 2020, who used virus medium, 5% humidity
*** UV LED high power UV germicidal 176 lamp (effective UV wavelength 260-285nm) without the titanium mesh plate (LEDi2, Houston, Tx) 50 cm from the UV source. At 50 cm the UVC power was measured by the manufacturer at 550 μW/cm2.
Further reading on the effect of heat: https://www.biorxiv.org/content/10.1101/2020.04.11.036855v2
Likely No. The method is efficient to sterilize the mask. Some N95 models can be reused after exposure to large doses of UV for sterilization, with little loss of filtration (using NaCl method, Viscusi et al., 2007, Viscusi et al., 2009, Lindsley et al., 2015) and strength. However, other models cannot because they lose 90% of their resistance and can lead to breaking during cough and infection. Fit factor can also be degraded after only 3 cycles (Fisher et al., 2020). Given it is not clear how to recognize which model is safe, it is better not to use that technique (Lindsley et al., 2015). Note that the sun does not contain the UVC necessary for sterilization, but only UVA and UVB. However, the sun can still be used to kill the virus.
Further reading: a review stating that is safe: https://osf.io/29z6u
Effect of the sun on the COVID: https://en.adioscorona.org/questions-reponses/2020-06-22-le-virus-r%C3%A9siste-il-aux-rayons-du-soleil.html#undefined and https://academic.oup.com/jid/advance-article/doi/10.1093/infdis/jiaa274/5841129
Certainly No. N95 and P100 respirators cannot be disinfected with alcohols such as isopropanol or ethanol because alcohols remove the electrostatic charge from the filtration media and substantially degrade its filtration capacity (Martin & Moyer, 2000 - cited from Lindsley et al., 2015, Viscusi et al., 2007, Lin et al., 2020, Liao et al., 2020). Fit factor is also significantly degraded as soon as the second cycle (Fisher et al., 2020). Surgical (Gauze and Spunlace masks) are also strongly degraded (Lin et al., 2020).
(for microwave-generated steam or microwave irradiation, see other sections)
Maybe no. Hydrophobic respirators placed in steam bags MSB X were partially sterilized (inactivated 99.9% of MS2 virus, which is less resistant than COVID-19 to heat) and retained their filtration efficiency after 3 cycles but no fit test, or resistance test was provided (Fisher et al., 2011). It is important to follow instructions carefully. Note that non-hydrophobic models were not tested, and that higher levels of sterilization (99.999%) is likely required given the likely infectiosity of the COVID-19.
(for steam bags, or microwave irradiation, see other sections)
On N95 respirators, a 1100W microwave steaming method inactivated between 99.99% (after 1 min) to 99.9999% (after 2 min) of the bacteriophage MS2 virus (Zulauf et al., 2020, see picture below), which is more resistant to heat than the COVID-19. Their method involves securing the mesh from a produce bag on top of a 17x7x7.5cm glass box with a rubber band, and adding 60mL of water. Fit factors were not affected by 20 cycles at 3 min.
A similar method involves placing the respirator on plastic containers with holes and water in the bottom and letting it in the microwave for 2 min at 1250W. This method deactivates >99.99% of infectious H1N1 virus (Heimbuch et al., 2011, see picture below, Gertsman et al., 2020), measured by the TCID50 dose.
In a review (Gertsman et al., 2020), 7 studies estimated that microwave-generated steam does not decrease filtration efficiency, fit factor or airflow resistance after 3 cycles of 2 min at 750W/ft3 (Bergman, 2010, 2011 and Viscusi, 2011, cited from Gertsman et al., 2020). However, sterilisations were made with a higher power (1100W-1250W).
Certainly Yes. On cloth masks, quilted masks, surgical masks and N95 respirators, a steaming method lasting approximately 13-15 minutes inactivated between 99.999% and 99.9999% of Staphylococcus aureus and the bacteriophage MS2 virus (Li et al., 2020). Those two organisms are thought to be more resistant to heat than the COVID-19 (as a comparison, dry heating it at 100°C for 15 minutes in an oven inactivated only 10-99.9% of it). The method included 8-10 minutes of heating and 5 minutes of steam in a rice-cooker-steamer. No deformation was seen on the masks. Filtration efficiency was preserved for 10 min over a beaker of boiling water for no more than 3 cycles (Liao et al., 2020) but also up to 5 cycles with no changes on fit factor .
A similar method involves placing the respirator on plastic containers with holes in a closed box with water in the bottom and letting it in the oven at 65°C for 30 min. This method was variable, deactivating between 99.9% and 99.9999% of infectious H1N1 virus (Heimbuch et al., 2011, see picture below). However, steam sterilisation at high temperatures (134°C) has been reported to cause mask deformation and failure on fit tests (ECDC Technical Report, 26 March 2020). This section is not about autoclaving.
Further reading: moist heat on N95 testing all aspects up to 10 cycles https://www.medrxiv.org/content/10.1101/2020.05.25.20112615v2
(for steam bags or microwave steaming, see other sections)
Likely No. Placing a N95 or P100 respirator directly on the revolving glass plate in the microwave and irradiating it for 1 minute each side at 750W/ft3 had been reported to cause deformation or significant decrease in filtration efficiency (Viscusi et al., 2007). Irradiation for 2 minutes on each side, without a paper towel, resulted in mask deformation and significant impairment on filtration efficiency. A later study showed that 6 out of 9 tested models of masks partially melted when being left 1 minute on each side at 750W-1100W.ft-3 (Viscusi et al., 2009). Please note that in the latter study, a paper towel was placed between the mask and the glass plate to prevent the elevated temperature of the glass plate from melting the mask, as intended by the author. It was reported that 45sec to 1 min was necessary to reduce the viral load by 99.9% (Gertsman et al., 2020).
Yes but it depends on the T°C. 99.9% of virus was inactivated after 30 min at 70°C on a mask, and 99.999% after 1 hour (Fisher et al., 2020). Respirators were left one hour in the oven and some models of respirators melted at 100°C and others saw their filtration efficiency decreased below 95% (Viscusi et al., 2009). None of the 9 tested models melted at 80 or 90°C or had filtration efficiency reduced below 95%. Dry heating at 80°C for an hour (Viscusi et al., 2007) or at 75-100°C for 30 min (Liao et al., 2020, ) did not significantly decrease the filtration efficiency below 95% or change breathability, strap elasticity or shape, even after 10 , 20 (Viscusi et al., 2007) to 50 cycles (Liao et al., 2020). However, another study found decreased fit factor after only 3 cycles although always above 100, which is still considered good fit (Fisher et al., 2020). A review (Gertsman et al., 2020) also concluded that low dry heat (<90°C) does not reduce filtration efficiency, fit or airflow resistance. Therefore, it seems safe to let respirators at 70°C for an hour for sterilization. Heating a surgical mask in a rice cooker for 3 minutes without adding water (149-164°C) was shown to retain the mask filtration (Lin et al., 2017). However, it is unclear whether this is enough to inactive 99.999% of the virus.
Likely Yes. Heating a surgical mask in a rice cooker for 3 minutes without adding water (149-164°C) was shown to retain the mask (gaze or spunlace) filtration (Lin et al., 2017). However, it is unclear whether this is enough to inactive 99.999% of the virus.Taiwan FDA also announced at a press conference recently that a surgical mask could be reused four to five times using a dry-heat method before disposal (Kan, The Liberty Times, 5 April 2020). To do this, place the surgical mask flat and directly in the insert pot of a rice cooker and, without adding water, turn on the rice cooker for 3 minutes followed by 5 minutes resting in it. This process did not compromise the Bacterial Filtration Efficiency of the mask, however, it is not clear how it affected fit factor, toxic waste, mask resistance or virus filtration efficacy. Dry-heating with a hairdryer or baking at 70°C for 30 minutes would not affect a surgical mask filtration (Song et al. 2020), at least for one round, but did after the second round, although still around 95% filtration. With the hairdryer, the authors put the mask in a plastic bag to avoid spreading the virus through airflow (but risk of melting the bag).
Certainly No. Bleach let significant off-gas and bleach smell on the N95 respirators (Viscusi et al., 2009), after 30 min submersion in a 0.6%-bleach solution (6%-chlorine) and overnight air-drying. Some find that it does not decrease the (NaCl) N95 filtration efficiency (Viscusi et al., 2007, Viscusi et al., 2009) but others that it does (for N95 and spunlace masks: Lin et al., 2020, Liao et al., 2020) and that it affected fit factor (Lin et al., 2020). It also destroyed gauze masks (Lin et al., 2020).
Yes. Two meta-analyses assessed the prevalence of comorbidities in 1576 and 1590 infected patients. The most prevalent comorbidities among infected patients were hypertension (21.1%), diabetes (9.7%), cardiovascular disease (8.4%) and respiratory system disease (1.5%, mainly chronic obstructive pulmonary disease COPD) (Yang et al., 2020, Guan et al., 2020). Pathologies associated with a severe form of the disease were a preexisting hypertension (odd ratio/OR: 1.58-2.36), cardiovascular disease (OR: 3.42), respiratory system disease (OR: 2.46) (Yang et al. 2020), and diabetes (OR 1.59) or malignancy (OR 3.50), after controlling for confounding factors (Guan et al., 2020).
A study (Ellinghaus et al, 2020) including 1980 patients with a severe manifestation of COVID-19 in seven Italian and Spanish hospitals, identified the genetic group 3p21.31 as a locus of genetic susceptibility in patients with COVID-19 with respiratory failure. This study also confirmed a potential association with ABO blood groups. In this cohort, a specific blood group analysis showed a higher risk in group A than in other blood groups, and a protective effect for group O compared to other blood groups.
Yes. The virus is in the blood given it is detected in blood samples (Wang et al., 2020). It can infect endothelial cells which line the blood vessels and cause endothelial inflammation (Varga et al., 2020) and more thrombosis. It can infect the eyes (Zhou et al., 2020). It can infect the brain and central nervous system (Conde et al., 2020) and trigger heart injuries (Huang et al. 2020). When it infects the brain, anosmia and ageusia can occur. Each of these last two symptoms have a high positive predictive value for SARS-CoV-2 infection (84.7% and 88.1%, respectively).
Brain infection: https://onlinelibrary.wiley.com/doi/full/10.1002/jmv.25728
More sources on brain infection: https://en.adioscorona.org/questions-reponses/2020-05-30-le-coronavirus-sars-cov-2-peut-il-affecter-notre-cerveau.html
The hypothesis of the Bradykinin storm: https://elifesciences.org/articles/59177
Likely Yes. 77 confirmed COVID-19 patients with acute respiratory distress syndrome were compared to 145 non-COVID-19 patients with acute respiratory distress syndrome (Helms et al., 2020). They developed significantly more thrombotic complications, mainly pulmonary embolisms (11.7% vs. 2.1%, a risk multiplied by ~6).
Yes. COVID-19 seems to induce reduced innate antiviral defenses (Interferon IFN I and III) coupled with exuberant inflammatory cytokine production (chimiokines and interleukines IL-1RA et IL-6), based on cell and animal models of COVID-19 infections, in addition to transcriptional and serum profiling of COVID-19 patients (Blanco-Melo et al., 2020). The results suggest an unique immune imbalance: low levels of interferons reduce a cell’s ability to limit viral replication, and the activation of less-specific immune responses promotes inflammation. Also, most lymphocytes T are decreased by the virus, including memory Th (Quin et al., 2020, Xu et al., 2020), with little change during convalescence (Zheng et al., 2020). That said, 99% of the people develop antibodies (potential immunity) against the virus after they recover (Wajnberg et al., 2020).
Possibly Yes. Even though asthma does not seem a risk factor regarding virus infection (Jin-jin Zhang et al., 2020), a correlation between asthma (and other respiratory conditions) with an increased severity of the patients with Covid19 has been reported (Zhang (H) et al., 2020).
Possibly Yes. Researchers  examined cardiac MRIs from 100 recovered COVID-19 patients aged 45 to 53 (77% of the COVID-19 patients had recovered at home). None of them had pre-existing cardiomyopathy. 78% were found with abnormal cardiac imagery and 60% with an ongoing myocardial inflammation, which was independent of preexisting conditions, severity and overall course of the acute illness, and the time from the original diagnosis.
Image from Singh et al. (2020) illustrating the mechanism:
Sars-CoV-2 comes from a bat virus. The closest bat virus is RaTG13, 96.2% identical to Sars-CoV-2. There is a common bat-virus ancestor to these two viruses . RaTG13 contaminated 4 miners in a cave of South China (Yunnan), who developed a serious pneumonia in 2012. Therefore, we know that the ancestor virus originated from that location.
The virus was not built in a laboratory by assembling other pieces of virus but evolved naturally by mutations and recombinations . The reason is that the mutations are numerous, small and made of sequences not present in any database of genomic banks of coronaviruses. The solutions found to bind to ACE2 receptors (or the polybasic cleavage site) are unique and therefore not a manipulation . There are different plausible scenarios. First, the ancestral virus could have evolved in a population of intermediary hosts, like the pangolins that were sold on the Yunnan seafood market in Wuhan, recombining with the Pangolin-CoV virus to gain its receptor-binding properties and then mutating a polybasic cleavage site, becoming highly dangerous for humans, before infecting the first human. However, no traces of SARS-CoV-2 were found in the animal samples from the market. Second, the ancestral virus could have directly infected a human, either in a Yunnan cave during sample collection, or during collection of bats to be sold on the market (or other opportunities). However, it is not known whether bats were sold on that market. The virus could have leaked from the Wuhan virology institute.
Certainly No. It was noticed that chunks of the COVID-19 sequence were similar to chunks of the HIV-1 virus (Pradhan et al., 2020). However, the article was withdrawn after it was made known to the authors that those chunks were also similar to chunks of bat coronaviruses.
Image from Bar-On et al. (2020) [see the article for references]:
100 nm = 0.1 μm
Likely Yes. While other viruses like flu, MERS or SARS do show a strong correlation between virus load and severity of symptoms, current research on covid-19 shows no strong consensus regarding this question. While Cereda D. et al (2020) finds no difference in viral load between symptomatic and asymptomatic patients (study with 5000 infected people), Yang Liu et al (2020). concludes that the mean viral load on severe cases was 60 times higher (study with 76 people).
Yes. And this virus is frequently mutating and you can follow the mutation data here.
Yes. Detected in January 2020 in Malaysia and quickly dominating globally in March, the D614G strain was found to be more infectious but not more deadly (Korber et al., 2000), confirmed by in vitro infectivity tests. The mutation actually happened in China or Germany (Korber et al., 2000).
Possibly. Detected in November 2020 in England, the B.1.1.7 strain may also originate from England. It was found to spread 70% quicker and not more deadly (Volz et al., 2020). The strain infects adolescents a little bit more, so it is possible that super-spreading events that they are more likely to assist explain the strain spread, rather than its properties. Indeed, the results come from a model and we are waiting for demonstrations coming from in vitro infectivity tests, or at least, higher viral loads.
Certainly Yes. A patient whose serum was frozen, was retrospectively tested positive to COVID-19 in Paris, France in December 2019 (Deslandes et al., 2020), although the first official case is currently on Jan 28. He presented typical COVID-19 symptoms, including the ground glass opacities on the lung images. First north-american patient is also on Jan 28.
- Spread history of the virus in the UK; the coronavirus has been imported to the UK more than 1,300 times — mostly from France and Spain (preprint): https://virological.org/t/preliminary-analysis-of-sars-cov-2-importation-establishment-of-uk-transmission-lineages/507
- Spread in the US could have started in December according to a model: https://www.medrxiv.org/content/10.1101/2020.07.06.20140285v1
Vous trouverez une réponse scientifiquement documentée pour la plupart des questions sur cet excellent document: https://www.infectiologie.com/UserFiles/File/groupe-prevention/covid-19/vaccins-covid-19-questions-et-reponses-spilf-24dec2020.pdf
We have several efficient vaccines, and most questions are answered here: https://en.adioscorona.org/themes-questions/vaccins.html
Two large independent studies  confirm the efficiency of the AstraZeneca/Oxford vaccine (ChAdOx1) and of the Pfizer/BiotNtech vaccine (BNT162b2). There is no difference of efficiency between younger than 65, older than 65 or older than 80, and no difference of efficiency between vaccine types. Efficiency at first dose was around 90%.
Likely yes. In United States, between December 2020 and Februrary 2021, through mRNA vaccine monitoring system, 35691 women were identified as pregnant. Among them 3958 accepted to enroll in a pregnancy registry. Injection-site pain was more frequently reported among pregnant women, headache, myalgia, chills and fever were less reported. Adverse pregnancy and neonatal outcomes in post-mRNA vaccinated pregnant women against covid and pregnant women from previous studies prior to covid-19 pandemic had similar incidences. No neonatal deaths were reported . Regarding lactating women, 13 human milk samples from 7 volunteers were collected between 4 to 48h after mRNA vaccination. None of the samples showed detectable levels of vaccine mRNA in the milk components. Vaccine-related mRNA does not seem to be transferred to the infant. Breastfeeding individual should not stop (may it be temporary of permanently) breastfeeding after receiving the COVID-19 mRNA-based vaccine .
Likely yes. In a randomized controlled trial (preliminary report ), with 11.320 randomized patients, including 2.104 in the Dexamethasone arm, the 28-day mortality in the Dexamethasone arm was significantly reduced compared to the control arm. The benefit was greater for the most severe patients, i.e. those receiving invasive ventilation with a reduction in mortality of 35% and 20% for those on non-invasive oxygen therapy. The benefit has not been demonstrated for non-oxygen-requesting patients.
This treatment also seems to reduce the length of hospital stay and reduce the risk of an unfavorable respiratory development for patients who are not severe on admission.
This Recovery study is methodologically robust and very well done.
Another study  conducted between 2013 and 2018 on ARDS (Acute Respiratory Distress Syndrome), reinforces this Recovery study. Indeed, in this randomized controlled trial, with a network of 17 intensive care units in university hospitals across Spain in patients with established moderate to severe ARDS, they recruited 277 patients and randomly assigned 139 patients in the dexamethasone group and 138 in the control group. At 60 days, 29 (21%) patients in the dexamethasone group and 50 (36%) patients in the control group had died. The proportion of adverse events did not differ significantly between the groups. It was found that early administration of dexamethasone could reduce the duration of mechanical ventilation and overall mortality in patients with established moderate to severe ARDS.
Dexamethasone binds with a high affinity to the same sites of the SARS-COV-2 Mpro than the Remdesivir, suggesting a specific COVID-19 action . Note that dexamethasone is a corticosteroid commonly used in intensive care so that it is likely that it also shows a non-specific effect against the COVID-19.
Further reading: Dexamethasone could increase fatal parasites for high doses: https://jamanetwork.com/journals/jama/fullarticle/2769100
Other corticosteroids do not seem to work or are harmful: https://www.sciencedirect.com/science/article/pii/S187140212030223X
Maybe yes. Three studies showed a positive effect but they had a very low level of evidence, since very few patients were included. Pending confirmation by the ongoing SOLID-C19 trial, preliminary results on Eculizumab have shown that it can be a key player in severe cases (Diurno et al., 2020). Treatment with 3 doses of Meplazumab IV at the start of pneumonia with SARS-CoV-2 could improve and accelerate the viral clearance, the severity of the disease, the radiological findings and the patient discharge time, but these results must be confirmed by a larger trial (Huijie et al., 2020). Siltuximab has shown normalization of CRP after 5 days of treatment (Gritti et al., 2020).
Certainly no. Two meta-analyses conclude that hydroxychloroquine (HCQ) has no effect [1,2], and can increase mortality when associated with azithromycin .
A double-blind randomized clinical trial showed a mild effect of hydroxychloroquine (HCQ) on patient’s recovery (Chen et al., 2020). Body temperature and cough recovery time shortened by 1 day, and pneumonia symptoms improved (using chest CT). Each group had 31 patients of similar demographics. The active group received 5 days of 400-mg-HCQ/day. Caution is required because the study is still not published after several months and it significantly deviated its main outcome than the one it registered.
However, another randomized clinical trial but open-label (Tang et al., 2020) found that HCQ did not decrease the number of days with the virus being detectable but improved clinical symptoms, possibly through anti-inflammatory effects. Dose was 200 to 800 mg-HCQ/day for 2-3 weeks and each group was 75 patients hospitalized with COVID-19. Adverse events were significantly increased in HCQ. Indeed, HCQ may trigger dangerous secondary effects (see next question).
A large RCT with open-labels  found no difference HCQ - control in mortality but patients in the hydroxychloroquine group were less likely to be discharged from the hospital alive within 28 days than those in the usual-care group (59.6% vs. 62.9%).
A large (96 032 patients) retrospective meta-analysis (Mehra et al., 2020) first concluded an absence of benefits for HCQ, chloroquine, or in combination with a macrolide (azithromycin), when started <48h after diagnosis. It was later retracted.
A randomized clinical trial in the monkey did not find an effect of HCQ or HCQ+azithromycin did not find an effect to decrease viral load or for prevention (Maisonnasse et al., 2020). Another study stands against HCQ (Barbosa, et al., 2020): a total of 63 patients were included with 32 in the HCQ arm. HCQ administration was associated with a need for escalation of respiratory support level compared to those who did not receive HCQ at 5 days (dose 200-400mg). The same findings were observed in a baseline-matched subgroup analysis. Absolute lymphocyte change in the HCQ group was no different than supportive care alone. HCQ use tended towards worsening neutrophil-to-lymphocyte ratio compared to supportive care alone as well as a higher risk for intubation. There were no benefits of HCQ on mortality, lymphopenia, or neutrophil-to-lymphocyte ratio improvement. This study however did not have a full randomization or double-blind.
For prevention of the virus, a double-blind randomized control trial assigned 821 people to take either HCQ or a placebo within 4 days of exposure to SARS-CoV-2 and HCQ did not protect those people more than the placebo (Boulware et al., 2020): however, they only reported symptomatic patients.
Another negative RCT: https://www.nejm.org/doi/full/10.1056/NEJMoa2019014
Another non-randomized positive result: https://www.ijidonline.com/article/S1201-9712(20)30534-8/fulltext
More on chloroquine (not hydroxychloroquine): https://www.jstage.jst.go.jp/article/bst/14/1/14_2020.01047/_pdf/-char/en
Maybe yes. 84 hospitalized covid-19 patients received daily doses of Hydroxychloroquine + Azithromycin (HQC+AZI) and an electro-cardiogram before treatment and around day 4 (Chorin et al., 2020). No torsade de pointe was observed but QTc was increased by 30 ms (p <0,001), with 11% of patients increasing by >500ms, being at risk for torsade de pointe and death. However, in the absence of a control group, not much can be concluded.
At normal dose, prolonged use is associated with a significant risk of retinopathy and blindness.
Certainly No. An official hospital page provides numbers seemingly allowing to compare the mortality rate in that hospital when treated by hydroxychloroquine and azithromycin (HCQ+AZI) with Marseille’s or the world mortality rate. Some media did that comparison (0.5% vs. 3% vs. 7%). However, Marseille’s or the world mortality rates are case fatality rates (proportion of death among confirmed cases). On the contrary, the specific hospital’s numbers reflect the infected fatality rate, because that specific hospital tests all people showing up no matter the reason and treats all positive cases with HCQ+AZI, no matter the severity. Therefore, those numbers cannot be compared.
Maybe yes. In a high-quality randomized clinical trial (Wang et al., 2020), 237 severe Covid-19 patients did not have their time to clinical improvement significantly reduced in the Remdesivir arm compared to the placebo arm: median of 21 vs. 23 days. 66% of patients taking Remdesivir also experienced side effects and 12% had to stop treatment early. However, it is important to wait for clinical studies in which treatment was started earlier. Indeed, a preliminary report (Beigel et al., 2020) for another high-quality clinical trial found a 31%-faster time to recovery in the Remdesivir arm with no secondary effects.
Further reading: a simulated two-arm study: https://www.medrxiv.org/content/10.1101/2020.05.02.20088559v1
Yes. At least 770 children have been confirmed to have caught the virus in China (Dong et al., 2020).
Yes. Children are less likely to be symptomatic or develop severe symptoms according to a systematic review (Ludvigsson, 2020). For example, in Dong et al. (2020), 13% were asymptomatic and 77% showed only mild to moderate symptoms, confirmed with a smaller study (Kelvin & Halperin, 2020). On the other side, some children subgroups seem more severely ill than adults (Cruz & Zeichner, 2020). There is a case-study of a unique child associating Kawasaki syndrome with COVID-19 (Jones et al., 2020).
Certainly yes. A Kawasaki-like disease in children is now called Paediatric Inflammatory Multisystem Syndrome (PIMS) (ECDC Rapid Risk Assessment, 15 May 2020). In Italy, the disease incidence increased 30-fold after the COVID-19 epidemic (10/month vs.0.3/month during the 5 years before) (Verdoni et al. 2020), and in France, 13-fold (Toubiana et al., 2020). Those children were older than usual, had a higher rate of cardiac involvement, showed macrophage activation syndrome and had marked gastrointestinal symptoms. In Paris, 59% of patients originated from sub-Saharan Africa or Caribbean islands. Most of them had contracted the COVID-19. The first published cases associating COVID-19 and Kawasaki disease were an infant in the US (Jones et al., 2020) and a small group of 28 children in the UK (without statistics) (Riphagen et al., 2020). The incidence of the syndrome for COVID-19-infected children could be as low as 0.16% (Morand et al., 2020). Out of the 176 cases reported worldwide, 5 children died (2.8% - ECDC Rapid Risk Assessment, 15 May 2020). Note that the association between COVID-19 and Kawasaki does not mean that one causes the other. Assuming the causal effect exists, the current risk of death from PIMS after catching the COVID-19 is 0.004% (4 for 100,000).
Certainly yes. Most of the data suggest no in-utero transmission (Peyronnet et al., 2020). However, a letter (Kimberlin & Stagno, 2020) mentions 3 neonates who may have been infected with SARS-CoV-2 in utero (Dong et al., 2020, Zeng et al, 2020, Zeng et al., 2020). Evidence is based on elevated IgM antibody values in blood drawn from the neonates following birth (IgM usually does not pass the placenta). None had positive PCR tests but three neonates from another study (Zeng et al., 2020) had positive PCR tests and positive chest tomographies (that have low false positive rate). Regarding the risk of fetal malformations, there is no data although we know that other coronaviruses do not cause it, but we must be careful.
Further reading: a clear evidence of one transmission of the virus from the mother to the child through the placenta: https://www.nature.com/articles/s41467-020-17436-6#Fig3
Certainly yes. [See more up-to-date info at Adios Corona]
Direct evidence of transmission from children to adults exists but is limited (Cao et al., 2020). In addition, even if asymptomatic, children do not show smaller viral loads than adults (Colson et al., 2020, Jones et al., 2020). Even, children <5yo show viral loads 10-100 higher in the nose than other children and adults .
There is evidence of fecal shedding in the stool for several weeks after diagnosis (Cai et al., 2020), leading to concern about fecal-oral transmission (Cruz & Zeichner, 2020). However, it is still not clear whether this RNA is infecting or not. Non-COVID-19 coronaviruses are detectable in respiratory secretions in a large percentage of healthy children (Heimdal, et al., 2019), and the extent to which this is also seen in COVID-19 is unclear but likely.
In this section, we characterize the basic epidemiological indicators: basic reproduction number, infection fatality rate, durations of latent, incubation and contagion periods, average onset-to-death time, and we compare some of those with influenza.
Most likely value: 0.66%-1.3% (China), 0.5% (USA).
The Infection Fatality Rate (IFR) strongly depends on the environment, health, healthcare access, age and strain of the virus.
In China, large-scale epidemiological data were corrected for under-ascertainment, censorship. Using the prevalence of PCR-confirmed cases in international residents repatriated from China, adjusted to the Chinese demographics, the IFR estimate was 0.66% (Verity et al., 2020).
The study of the case of the Diamond Princess cruise ship (Russell et al., 2020) estimated that the Chinese IFR was 1.3% (0.38–3.6) after correcting for the fact that most people on the ship were older than the average Chinese population (stratification adjustment). Adjusting to the US population age yields a 0.5% rate (Faust & del Rio, 2020).
Jung SM, Akhmetzhanov AR, Hayashi K, et al. (2020) Real-time estimation of the risk of death from novel coronavirus (COVID-19) infection: inference using exported cases. J Clin Med.;9(2):E523
Covid-19 Antibody seroprevalence in Santa Clara county, California (2020) Bendavid et al. https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v1
Likely No. A peer-reviewed viewpoint argues that the number of US-CDC’s reported influenza deaths cannot be compared with the number of reported covid-19 deaths because the first is estimated through a model adjusting for massive underreporting, while the second corresponds to confirmed deaths (Faust & del Rio, 2020). If one compares the number of deaths reported under the same conditions in the USA, the COVID-19 weekly death number is 20 times higher than the influenza death number.
Likely 18 days. The median days from symptom onset to (only 17) deaths were 14 (range 6-41) days in one study , and tended to be shorter among people of 70 year old or above (11.5 [range 6-19] days) than those with ages below 70 year old (20 days [range 10-41])
Another larger study  estimated that time to be 20 days.
Certainly 5 days. A meta-analysis (5 studies) estimated the incubation period at 5 days with the 95% confidence interval between 4.8 and 5.4 . The incubation period is the time between infection and the onset of symptoms. Estimations of average incubation periods ranged from 4.8 days to 6.5 days (median 5.2, see sources estimates in the table of this page / ). From travel data, 97.5% of those who develop symptoms will do so within 11.5 days, 90% before 10 days and 99% before 14 days , which grants the duration of the quarantine period to 14 days . Modeling of 198 chains of transmission in China concluded to 7 days .
Possibly 9 days total and 5.5 days from contagious to isolation. The infectious period is the time during which the disease is contagious. It starts after the latent period, when contagiousness reaches 50% of its maximum, followed by a pre-symptomatic period of infectiousness, itself ended by the onset of symptoms. Contagiousness continues then and is maximal for a few days .
Contagiousness estimated from cell culture infection was zero when tested 10 days after symptom onset but RNA could still be detected 22 days after . End of contagion (defined as contagiousness on cell cultures at 50%) happens on average 7 days after symptom onset , which itself occurs on average 2 days after the latent period , for a total infectious period of 9 days. This is the most likely value. Viral charge from throat swabs decreases from maximum at symptom onset to 50% of maximum 3-5 days after onset (infectious period of 5.5 days) [2,8,9] but contagiosity of the swabs is not demonstrated (RT-PCR). Indeed, viral loads smaller than 50% of the peak could still be infectious. A similar method concluded to 7 days with the same lack of contagion demonstration . It is difficult to estimate the infectious period when it is estimated through models because they often report the time from onset to diagnosis or isolation (quarantined or hospitalized). It is what we investigate now. From the modeling of China’s data, the infectious period was 13.5 days  or 3.5 days . From real data, the onset to isolation time was estimated to 2.9 , which is an infectious period of 4.9 after adding pre-symptomatic contagion and to 4.6 , which is an infectious period of 6.6 days. The median of these studies is 5.6 days.
Further reading: This article, based on 100 COVID-19 patients and 2,761 people who had close contact with them, indicates that the risk of transmitting the virus is highest in the 5 days following the onset of symptoms and in the days preceding the onset of symptoms.
Certainly 3. The reproduction number is the average expected number of cases directly generated by an individual infection, at the start of an epidemic, when everybody is susceptible. A meta-analysis (7 studies, mostly based on China) estimates the basic reproduction number to be 3.15 with the 95% confidence interval between 2.4 and 3.9 , and another (13 studies in China) to be 2.8 (median) to 3.3 (mean) .
Individual studies: https://www.nejm.org/doi/10.1056/NEJMoa2001316
UK Re estimate to 2.6, prior to lockdown: Jarvis et al., 2020
USA R0 estimates (states with R0 from 1 to 6.7): https://www.medrxiv.org/content/10.1101/2020.05.17.20104653v3
Transforming growth rate to R0: R0 = (growth rate + 1/latent period)(growth rate + 1/infectious period)/(1/infectious period * 1/latent period): https://www.sciencedirect.com/science/article/pii/S2468042719300491?via%3Dihub
In october 2020, that probability is really low (0.27% of contaminated persons in France, 0.55% in Paris). On Oct 17, the probability that nobody is infected in a group of 100 random people is 89% in France, 78% in Paris, 83% in Toulouse, 79% in Marseille and Lyon.
See current stats here: How many infectious people in my street?
Likely 3 days. Once a person is infected, it takes a period of time known as the latent period before they are able to transmit the virus (technically, time to reach 50% of maximum contagiousness). Infectiousness was estimated to start 2.3 days before symptom onset , itself assumed to happen 5 days after infection, for a latent period of 2.7 days. Another model estimated it at 3.7 days . The median latent time is therefore ~3 days followed by ~4 days of close to maximal infectiousness .
Most likely, >20% of the people are completely asymptomatic . This is the result of a meta-analysis of 94 studies but the results are very variable across individual studies. This is because the meta-analysis aggregates data from hospitals and from everyday life (serological studies) which will strongly differ in those proportions. Indeed, in Vo’, 42.5% are asymptomatic , 31% in Lombardy , or 17% in the young people of a French high school .
In Vietnam (approximately half of an n=30-sample was asymptomatic): https://academic.oup.com/cid/advance-article/doi/10.1093/cid/ciaa711/5851471
According to a model , 46% of contaminations come from presymptomatic individuals (before showing symptoms), 38% from symptomatic, 10% from asymptomatic (who never show symptoms), and 6% from the environment. Results on the last two routes are speculative. Presymptomatic transmissions alone are almost sufficient to sustain epidemic growth.
Infected people can shed the virus via aerosol droplets 2.5 days before they start developing symptoms (pre-symptomatic transmissions). A meta-analysis (6 studies) estimates the pre-symptomatic infection rate is 46% with the 95% CI ranging from 18.4% to 73.6% .
Pre-symptomatic transmission accounted for 48% of all transmissions of a Singaporean cluster and 62% of the Tianjin cluster. From: https://www.medrxiv.org/content/10.1101/2020.03.05.20031815v1
And 50% of transmissions in Wuhan: https://www.nature.com/articles/s41591-020-0869-5?fbclid=IwAR2JVhkvotys2G1weOJ2wtN2gKegSPfaNcaqd-8Y_79UMMSmc7oOswwx0bo#Fig1
It is explained by the fact that asymptomatic people have the same viral load than symptomatic ones: https://www.nejm.org/doi/full/10.1056/NEJMc2001737
Estimate of 44% of pre-symptomatic transmission in China: He et al., 2020
Following the cases of the Diamond Princess cruise, the proportion of pre and asymptomatic people was 51.4%: https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2020.25.12.2000256?utm_source=sootoday.com&utm_campaign=sootoday.com&utm_medium=referral
Yes. Pre-symptomatic transmissions exist (infections at a stage where patients are asymptomatic - see difference between latent and incubation periods) and account for a large proportion of infections (see question about the proportion of asymptomatic transmissions). Transmissions from people who are and will stay filly asymptomatic are also documented.
Real cases of an asymptomatic transmission: https://jamanetwork.com/journals/jama/fullarticle/2762028
Likely yes. 99% of the people develop antibodies (potential immunity) against the virus after they recover (Wajnberg et al., 2020). The immune response could be slow: some study volunteers did not produce detectable antibodies until one month after they first started feeling ill.
Yes. There is good converging evidence through case studies, with at least 15 documented cases where one person gets infected, recovers and gets reinfected with a different strain of virus. For example, a patient who tested positive for coronavirus SARS-CoV-2 and 4.5 months after the first infection. The differences between the viral RNA sequences collected during the two infectious episodes confirm that these are two successive independent infections (To et al., 2020). A 25-year-old patient (Tillett et al., 2020) developed a severe form of COVID-19 (with hospitalization and need for continuous oxygen supply) one month after the end of symptoms associated with a first infection that did not require hospitalization. Differences in viral RNA sequences between the two infectious episodes confirm that this is a reinfection and not a resurgence of the first infection.
Most people react the same way to the infection and reinfection.
A 46-yo patient was tested positive for SARS- CoV-2 RNA test 5 days after a negative test for SARS-CoV-2 (Chen et al., 2020). A 68-yo patient was tested positive for SARS-CoV2 9 days after being discharged from hospital (after 2 consecutive negative tests), that same patient had a 3rd resurgence of the virus after the 2nd hospital discharge (after 4 consecutive negative tests), suggesting the possible reactivation of the virus (Cao et al., 2020). Five out of 55 patients (9%) suffering from a COVID-19 pneumonia tested after hospital discharge (Ye et al., 2020), presented a reactivation of the infection with positive RT-PCR tests, 4 to 19 days after the last negative RT-PCR. Four of them presented some of the classical COVID-19 symptoms. Four other patients considered recovered (RT-PCR negative) showed again positive in COVID-19 tests, 5 to 13 days after hospital discharge (Lan et al. 2020).
Some of the results above have to be considered carefully in light of high fluctuation of the results of swab tests in the later phases of COVID-19 (Wölfel et al., 2020). Importantly, a positive RT-PCR does not mean that the patient is infectious or sick. Indeed, PCR-positivity was detected for two or more weeks after symptom resolution (20% of the people) and up to 28 days (Wajnberg et al., 2020).
See more literature at Adios Corona: https://en.adioscorona.org/questions-reponses/2020-09-09-attraper-covid-2e-fois-contamine.html?fbclid=IwAR0s4CwNbMBUpqkCjmHe0l3uSWdkQWMj0UFuj1DnPiFnxrZ6qD8dl53wUQI
Reinfection in macaques: https://science.sciencemag.org/content/sci/early/2020/05/19/science.abc4776.full.pdf
Maybe Yes. Evidence suggests that the virus can reactivate in patients discharged from the hospital (see above). However, one needs to be careful because measures of viral RNA rather than infectious virus were taken so it is not demonstrated that they could spread the virus.
The critical proportion of contaminated people needed to reach herd immunity can be expressed as 1 - 1/R0, which is 77% for a likely R0 of 3.
Geneva, Switzerland, May 2020: Certainly no. In a representative population of Geneva (n= 1335), the prevalence of anti-SARS-CoV-2-IgG antibodies was less than 10%. “In the first week, we estimated a seroprevalence of 3.1%. This increased to 6.1% in the second, and to 9.7% in the third week”. Assuming that the presence of IgG antibodies is at least in the short-term associated with immunity, these results highlight that the epidemic is far from burning out simply due to herd immunity (Stringhini et al., 2020).
France, 11 May 2020: Likely no. Using an epidemiological model, an estimated 4.4% of the population (9,9% in Paris greater area) may have been contaminated (Salje et al., 2020).
Spain, 1 May 2020: only 5% https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)31483-5/fulltext
A model for UK and Italy’s immunity: https://www.medrxiv.org/content/10.1101/2020.03.24.20042291v1.full.pdf
A list of serological studies: https://docs.google.com/spreadsheets/d/17Tf1Ln9VuE5ovpnhLRBJH-33L5KRaiB3NhvaiF3hWC0/edit#gid=0
Prevalence in the USA, using a model to correct bias sampling: http://www.cireqmontreal.com/wp-content/uploads/cahiers/08-2020-cah.pdf?utm_source=M%C3%A9dias&utm_campaign=f293fa85dd-EMAIL_CAMPAIGN_2020_04_30_08_02&utm_medium=email&utm_term=0_0ea13f21a3-f293fa85dd-332389590
Maybe Yes. The virus binds to a target cell (see Image below) by deploying viral proteins (red) to connect to ACE2 proteins (blue) on the target’s surface (Letko et al., 2020). Certain populations and individuals may be better protected against the virus than others due to genetic difference [Copied from Iwasaki & Grubaugh, 2020].
Credit: Juan Gaertner/SPL
Maybe yes. Japan, like many other countries including China, Korea, India, and the Russian Federation have mandatory childhood BCG vaccines against Tuberculosis. These countries have so far a relatively low per capita death rate from COVID19 compared to countries that have no mandatory BCG vaccines (US, Spain, France, Italy, Netherland) . What further distinguishes Japan is that the BCG vaccine strain used is one of the original strains. This association between BCG vaccination and apparent low COVID19 incidence in Japan has spurred the idea that these two things may be linked . It is known that BCG provides partial protection against several viral respiratory diseases but it is not known whether it applies to the COVID-19  thanks to trained immunity .
Unknown but likely 1 copy of virus gives a 1% infection probability and 7-70 copies gives a 50% probability. The actual dose is unknown but we can make estimates. We define the infectious dose as the TCID50 or the number of virus copies that is enough to contaminate 50% of the people that are exposed to it (HID 50). A group (Augenbraun et al., 2020) estimated the infectious dose conservatively from influenza and SARS-CoV at 1 copy for a 1% chance to be infected. With a dose-response model, it corresponds to a HID50 of ~70 virus copies. Another estimate is based on the fact that the HID 50 is ~0.7 TCID50 for the common cold, 1 TCID50 for the adenovirus (Couch et al., 1966) and 0.6 to 3 TCID50 for the influenza (using non-immunized patients exposed by aerosol inoculation; Nikitin et al., 2014). Therefore, one can reasonably expect the infectious dose for the COVID-19 to be around ~1 TCID50 for aerosolized virus, which is equivalent to ~0.7 PFU (Plaque-Forming Units). The particle-to-PFU ratio can be in the range of 10:1 to 100:1 (Fonville et al., 2020), which translates the HID50 to 7-70 virus copies and the HID1 to 0.1-1 copy. The TCID50 calculated from swab viral loads is approximately 7.5 log10 RNA copies . However, we do not know how the TCID50 from the swab relates to the TCID50 for the aerosolized virus. Using a different method, an estimate of the upper limit for infectious dose was 300 .
Maybe >6% over 6 months. A good-quality model (Augenbraun et al., 2020) estimated that probability at 6% over 6 months (at 1h/week in large groceries stores), using a surgical mask each time. However, this is only taking into account the airborne risk and not the risk from close contacts, face-touching and fomite contaminations.
A risk visualization depending on the situation: https://www.nationalgeographic.com/science/2020/08/how-to-measure-risk-airborne-coronavirus-your-office-classroom-bus-ride-cvd/
Based on the Jimenez’s model: https://docs.google.com/spreadsheets/d/16K1OQkLD4BjgBdO8ePj6ytf-RpPMlJ6aXFg3PrIQBbQ/edit#gid=519189277
Certainly Yes. Cross-protective immunity, in the case of COVID-19, is referring to the protection against SARS-CoV-2 infection due to the pre-existing adaptive immunity developed from the past exposure to another coronavirus(es) [adapted from 1,2]. Early evidence suggests that cross-protective immunity may be present against COVID-19 in the human population . Using multiple experimental approaches, the SARS-CoV-2-specific immune response (CD4+ T cells) observed in blood samples from recovered COVID-19 patients (n=10, 100%) were also detected in healthy individuals (n=11, 40-60%), despite at significantly lower levels. Past exposure to two human coronaviruses known to cause seasonal ‘common cold’ upper-respiratory tract infections was found in these healthy individuals, indicating some potential for pre-existing immunity in the human population. The authors acknowledged the limitations of the study including low sample size, use of mild-to-moderate cases only, and the lack of pre-COVID-19 samples for comparison.
Further reading: confirmation of 50% cross-immunity in another study: https://www.nature.com/articles/s41586-020-2550-z
But the opposite in children: https://www.medrxiv.org/content/10.1101/2020.06.29.20142596v1
Likely Yes. Not necessarily due to the smoke, but due to the normal breathing and exhalation of the smoker (the smoke makes the breath visible): all other things being equal, smoking and not-smoking should equally spread the virus. Indeed, COVID-19 virus is primarily transmitted between people through respiratory droplets and contact routes. [2-7] In an analysis of 75,465 COVID-19 cases in China, airborne transmission was not reported." https://www.who.int/news-room/commentaries/detail/modes-of-transmission-of-virus-causing-covid-19-implications-for-ipc-precaution-recommendations
==> the main routes remain respiratory droplets and contact routes, so as long as the person is breathing, the risk should be similar to a non-smoker. The act of smoking might even increase the risk of contamination from the hands touching a (shared or not) lighter, a cigarette, then bringing it to the mouth
To dig deeper, there's 2 other questions to look into:
1) "Does secondhand smoke increase the spread?" may be comparable to "Is covid-19 airbone (ie: can it remain in the air for long periods of time and be transmitted over distances greater than 1m)?"
==> Rather than asking if the virus can "piggyback" on smoke, it's asking if it can move similarly to smoke. If yes, you could compare the movement of exhaled smoke to possible spread of the virus.
2) "Does secondhand smoke decrease the spread?" could be similar to "What if freshly exhaled smoke from a cigarette could reduce or deactivate the virus coming out of a person?" ==> Now, even if the answer is yes, the risk-benefit ratios from other measures are probably better...
1) Is covid-19 airbone: Possibly yes
"Assuming SARS-CoV-2 virions are contained in submicron aerosols, as is the case for influenza virus, a good comparison is exhaled cigarette smoke, which also contains submicron particles and will likely follow comparable flows and dilution patterns. The distance from a smoker at which one smells cigarette smoke indicates the distance in those surroundings at which one could inhale infectious aerosols. In an enclosed room with asymptomatic individuals, infectious aerosol concentrations can increase over time." Prather et al., 2020
2) Does exhaled smoke from cigarettes reduce or deactivate the virus in aerosols?
For the sake of argument, let’s just focus on the immediate effect of the smoke inhaled and directly exhaled on the virus, which usually last only a few seconds
Temperature wise, it seems like the answer is no.
The Exhaled Breath Temperature while smoking reported in some studies are usually below 37°C Quigley Jr. et al., 1965
According to this other paper you would need 90, 60 and 30 min exposure at 56, 67, and 75 degrees C (respectively) to make it non-infectious. Duan et al., 2003
What if one of the many components could do it in a few seconds? No papers looked into that as far as we know. Anyway it's certainly not a sustainable nor healthy solution to stop the spread vs wearing a mask!
Certainly yes. In closed environments the risk of transmission of the SARS-CoV-2 coronavirus is 18.7 times greater than in the open air. 
A risk visualization depending on the situation: https://www.nationalgeographic.com/science/2020/08/how-to-measure-risk-airborne-coronavirus-your-office-classroom-bus-ride-cvd/
Based on the Jimenez’s model: https://docs.google.com/spreadsheets/d/16K1OQkLD4BjgBdO8ePj6ytf-RpPMlJ6aXFg3PrIQBbQ/edit#gid=519189277
Further reading: Ranking the efficiency of crisis management actions: https://www.nature.com/articles/s41562-020-01009-0.pdf
Likely yes. Most infected people are close to peak infectiousness for about 4 days, beginning ~3 days after being exposed. A model estimated the efficiency of a 4-days-work/10-days-lockdown . The strategy allows most of those infected during work days to reach maximal infectiousness during lockdown. It reduces the contact time between people by 70% and the reproduction number below 1, while saving 40% of the economy. In a related staggered strategy, two groups alternate their work period on different weeks, allowing for continuous minimal activity without affecting the disease dynamics (see figure below).
The model depends on assumptions about a latent period of 3 days (that seems to be correct) and an infectious period of 4 days (there is considerable variation in estimates of that period but median estimates of contagious-to-isolation period is closer to 5.5 days). 90% of symptomatic people will develop symptoms in 10 days from infection . In modelling studies, lockdown was estimated to decrease the reproduction number by 84%  and 85 to 90% . Applied to the R during work days, the number of days of work allowed in 14 days to keep average R<1 follows the red line in the figure below (which shows the number of allowed work days to keep average R<1). In conclusion, the 4-days-work/10-days-lockdown strategy is efficient, as long as the R during work days is <3 (which is the most likely R0 before interventions). Lower R during work days (with masks for example) could possibly counteract the likely longer contagious period than the one assumed.
Interestingly, such a strategy can be followed at the individual level but should be avoided if one is a vulnerable person given it only prevents infecting other people once infected.
Interestingly, the model is robust to change in the parameters. It does not include a parameter for asymptomatic people, but assuming a proportion of asymptomatic people of 20% (most likely value), the model can take into account that proportion with a leakage parameter at 30% (20% asymptomatic, 10% non-compliance) and it does not change the conclusion.
Further reading: a similar model with same conclusions: https://arxiv.org/abs/2006.06409
Resources and current implementations of the strategy: https://cyclicexitstrategy.org/
Yes. Lockdowns are also called shelter-in-place or home isolation. At least one model  with numerous assumptions relative to the Italy outbreak concluded that only a strict lockdown could contain the outbreak. Interestingly, the number of infections was only marginally influenced by the proportion of asymptomatic cases. Lockdown decreased the reproduction number from 3.3 to 0.5 (-84%). However, the study failed to define conditions for the outbreak to be declared contained. Another model  estimated that the French lockdown prevented >61,000 deaths during its first month. They took into account regional differences in the model parameters and hospital capabilities. It was shown that during lockdown, people in Shanghai and Wuhan cut their encounters with others from 15–20 to ~2 per day , which was enough to bring the epidemic under control in the two cities, according to their model. In the UK, contacts decreased from 10.8 to 2.8, producing a 74%-R0-reduction . Lockdown decreased the reproduction number by 85 to 90% . Shutdown orders prevented about 60 million novel coronavirus infections in the USA and 285 million in China . The shutdowns saved ~3.1 million lives in 11 European countries and dropped infection rates by ~82%, according to an estimation of actual policies through reduced-form econometric methods .
Further reading: Lockdownd and alternative scenarios: http://www.imperial.ac.uk/mrc-global-infectious-disease-analysis/covid-19/report-9-impact-of-npis-on-covid-19/
A Science-published model for China: https://science.sciencemag.org/content/early/2020/04/07/science.abb4557
Transforming growth rate to R0 / R0 = (growth rate + 1/latent period)(growth rate + 1/infectious period)/(1/infectious period * 1/latent period): https://www.sciencedirect.com/science/article/pii/S2468042719300491?via%3Dihub
Certainly Yes. The benefit of mass face mask adoption on reducing the reproduction number (R0) and epidemic growth of COVID-19 has been demonstrated by several models:
Reduction of R0/Rt: By using on a quantitative transmission model (Tian et al., 2020), a conservative assessment of COVID-19 with R0 of 2.4 (Ferguson et al., 2020) when wearing no mask, 100 cases at the start of a month would become 31,280 cases by the end of the month vs. only 584 cases when there is a 50% adoption in the population (R0 = 1.35, 50% of mask efficacy level) (Howard et al., 2020, Preprint). If mass mask-wearing is adopted regardless of the presence of symptoms, it alone can bring a high R0 (2.2-4) down to below 1 even if the effectiveness of the masks is only 50-75% (Stutt et al., 2020).
Reduction of epidemic growth: The adoption of universal mask-wearing, together with other crisis management measures, can substantially control the epidemic growth of COVID-19. Mask-wearing alone was also deemed to reduce the risk of SARS virus transmission by 68% (Jefferson et al., 2011, cited from Salvi, 2020), and with hand washing, wearing gloves and protective gear by 91%. Together with lockdown, 50% of adoption (mask at 50% efficacy) can prevent any exponential growth of infected cases, while 100% of adoption can flatten the curve and stop the occurrence of future waves (Stutt et al., 2020). Even if there is a delay in adoption, a universal adoption of face masks can stop the occurrence of further COVID-19 epidemic waves from the point of adoption (Stutt et al., 2020). Together with contact tracing (60% efficiency within a 4-day time frame), 70% of adoption can flatten the epidemic growth in hard-hit countries (Tian et al., 2020).
*Note: efficacy is likely to be referring to the bacterial filtration efficiency tested using aerosol Staphylococcus aureus with a size at 3.0 µm ± 0.3 µm (ASTM International, 2001, cited from Tian et al., 2020)
A risk assessment model for an influenza epidemic demonstrated that if people do not wear a mask, around 35% of the people will catch influenza, if 50% of the people wear a mask, or if the efficacy of the mask is 50%, the prevalence of infection will reduce by 50%, and if 80% of the people wear the mask and adhere to it, the risk of transmission will be negligible. Please note that influenza is 2 times less contagious than COVID-19, and thus it is reasonable to assume the impact to be higher for COVID-19: Yan et al. 2019, cited from Salvi, 2020
Evidence review: Applying precautionary principle on wearing face masks for policy making: https://www.bmj.com/content/369/bmj.m1435.full
Maybe yes. But not enough. A model based on actual interactions counts  suggests that, in Shanghai, school closures alone would not have stopped the epidemic — but lower the peak incidence by 50%. An estimation of actual policies through reduced-form econometric methods only found a marginal effect of school closures on the epidemic growth .
Further reading: -58% mortality after school closure in the US: https://jamanetwork.com/journals/jama/fullarticle/2769034 but see also https://jamanetwork.com/journals/jama/fullarticle/2769033
Certainly Yes. At the individual level, the virus is sensitive to heat, humidity and UV. At the population level, it is difficult to disentangle covarying parameters: humidity and temperatures covary, and the smaller number of cases in warmer countries can also be explained by the fact that they did far less testing in the beginning . The pandemics started in the Northern hemisphere in cold months and the increasing number of cases covaried with the increase in temperatures and humidity.
However, full-year global data exists now and UV light has the strongest effect (, 7% change in daily growth rate by SD increase): the effect is likely causal given that panel analysis was used, and it accounts for co-varying parameters like temperatures, specific and relative humidities (but absolute humidity was not tested). Another model estimated that <17% of the growth rate variance could be explained by 4 parameters (UV, relative humidity, proportion of elderly and T°C) and UV light was the strongest predictor, reversing the effect of heat (which then increased transmission) . Contrary to most studies, these two studies controlled for the lag between the climate event and the propagation change.
More partial data with less robust methods  found that a 1°C-increase in temperature lowered the reproduction number (R) by 0.0225 (0.7% of average R0=3). For absolute humidity, each additional g/kg decreased R by ~0.5% , and relative humidity can play a role too (0.5% R-decrease by each additional point ). With another set of partial data and non-robust methods, there seems to be an ideal absolute humidity range between 4 and 10 g / m3 at which most of COVID-19 cases occur , with the ideal range for temperatures being 2°C-18°C .
The mechanism is unclear, given the complex intrication of the parameters: the aerosol and surface virus survival is influenced by temperature, humidity, and UV light, but social gatherings are also influenced by the weather, and differently across countries.
Average T°C, relative humidity and absolute humidity, worldwide.
Further reading (in French): https://www.santelog.com/actualites/covid-19-sars-cov-2-sera-t-il-saisonnier-comme-les-autres-coronavirus)
Being more than 1 meter away from each other decreases the risk of catching acute respiratory syndromes (13%) by 80% (to 2.5%), according to a recent meta-analysis .
 Chu et al., 2020
See also https://www.medrxiv.org/content/10.1101/2020.04.08.20058578v4?fbclid=IwAR15TNUQdUK2RLJ3ZRahUTN_FoJwOy3j1wNayEz_tqrdVIdULqk_zDihaf0 from https://www.technologynetworks.com/tn/news/researchers-have-discovered-a-strong-correlation-between-severe-vitamin-d-deficiency-and-mortality-334567?fbclid=IwAR2ROTvkLuJ4vDQitro776p3WKIpFjvN6Y4S3z8bPMzw6dd1_1-wZEnN_sI
Careful with MDPI’s: https://www.mdpi.com/2072-6643/12/4/988
Likely, the probability to be infected is divided by 4 or 5 in smokers Miyara&Tubach_qeios, but once infected, smokers die x2.5 more than non-smokers: https://www.journalofinfection.com/article/S0163-4453(20)30234-6/pdf
Likely, see articles in: https://www.latimes.com/science/story/2020-06-26/inside-the-body-the-coronavirus-is-even-more-sinister-than-scientists-had-realized?fbclid=IwAR19ZD4uFoX3Ii4uMdBrFO0Ez3r3f0Ylsh1m7o4OvVvx3zkOLlBS9PFFz1I
Seemingly important contamination route: the eyes (contagion x100 compared to Sars-Cov): https://www.sciencedirect.com/science/article/pii/S2213260020301934
Stool: seems yes: See
But not in: Wölfel et al., 2020
There is now enough evidence and it is described in details here (FAQ from a group of scientists in charge of answering that question):
And in the older https://www.nap.edu/read/25769/chapter/1 and Santarpia 2020 https://www.medrxiv.org/content/10.1101/2020.03.23.20039446v2 or Tang 2020 https://www.sciencedirect.com/science/article/pii/S0160412020319942
Contagiousness in air samples: https://www.medrxiv.org/content/10.1101/2020.03.09.20033217v1.full.pdf
Morowska and Cao, 2020: https://www.sciencedirect.com/science/article/pii/S016041202031254X
Complex case: see oppinion https://www.nature.com/articles/d41586-020-00974-w
Careful with Zhang et al. in PNAS: https://metrics.stanford.edu/PNAS%20retraction%20request%20LoE%20061820
"Assuming SARS-CoV-2 virions are contained in submicron aerosols, as is the case for influenza virus, a good comparison is exhaled cigarette smoke, which also contains submicron particles and will likely follow comparable flows and dilution patterns. The distance from a smoker at which one smells cigarette smoke indicates the distance in those surroundings at which one could inhale infectious aerosols. In an enclosed room with asymptomatic individuals, infectious aerosol concentrations can increase over time." Prather et al., 2020
CDC sourced report on N95’s reuse: https://www.cdc.gov/coronavirus/2019-ncov/hcp/ppe-strategy/decontamination-reuse-respirators.html
Ontario’s Public Health report (with limited info on surgical mask): https://www.publichealthontario.ca/-/media/documents/ncov/covid-wwksf/what-we-know-reuse-of-personal-protective-equipment.pdf?la=en
For a comparison with France including cultural differences: https://journals.sagepub.com/.../10.1177/0275074020942445
Cohort study of magnesium, vitamin D and B12 for COVID-19: https://www.medrxiv.org/content/10.1101/2020.06.01.20112334v1?fbclid=IwAR02NH19ohN9tpd27c5yBC8-J78QuI-Fa4YclogMIuYCL372Df_8ttX5wB4
Not decreasing lung inflammation in H1N1 infection using Vitamin C and Ginseng in mice: https://www.ncbi.nlm.nih.gov/pubmed/26898166
Model in Science: https://science.sciencemag.org/content/early/2020/04/09/science.abb6936
Hellewell J, Abbott S, Gimma A, et al (2020) Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts. Lancet Glob Health.;8(4):e488–e496
This study concludes that no (they found two cases in samples taken for other reasons, in Feb 21 and 23, on 2888 samples: https://jamanetwork.com/journals/jama/article-abstract/2764364
A well-furnished review on various topics: Zimmerman & Curtis, 2020
Another set of interesting reviewed and sources info on transmission: https://www.worldometers.info/coronavirus/transmission/
Everything below thanks to Pr ANNE-CLAUDE CREMIEUX:
See https://www.medrxiv.org/content/10.1101/2020.03.31.20038935v1 Etude Rétrospective Monocentrique Chinoise évaluant l’impact de l'utilisation des ARA2 / IEC sur la gravité du COVID-19 126 patients COVID-19 ayant une hypertension dont 43 sous ARA2/IEC. Le groupe ARA2 / IEC avait des concentrations plus faibles de CRP (p =0,049) et de procalcitonine (PCT, p = 0,008) par rapport au groupe recevant d’autres antihypertenseur. Une proportion plus faible mais non significative de patients critiques (9,3% vs 22,9%; p = 0,061) et un taux de mortalité plus faible (4,7% vs 13,3%; p = 0,216) ont été observés dans le groupe ARA2 / IEC que dans le groupe non- ARA2 / IEC. See also: https://www.tandfonline.com/doi/full/10.1080/22221751.2020.1746200 Etude rétrospective monocentrique chinoise évaluant l’impact de l'utilisation des ARA2 / IEC sur la gravité du COVID-19 - 42 patients COVID-19 ayant une hypertension dont 17 sous ARA2/IEC. Dans le Groupe ARA2 / IEC seuls 4 patients (23,5%) ont été classés comme graves et aucun patient n'est décédé 12 patients dans le groupe non traité par ARA2 / IEC (48%) ont été classés dans la catégorie grave et un patient est décédé. Deux Etudes avec beaucoup de limites méthodologiques car nombreux biais et facteurs de confusions, mais qui seraient plutôt en faveur de ne pas interrompre les IEC/ARA2
In spite of all the "research shows", "scientists found" seen online, I can't find many published studies on that matter besides  and  (conference abstract by an industry group) that looked into influenza-like illness measured by wearables.
Specifically on covid-19, all I have is an anecdote from one guy who happened to be tested positive after his "readiness score" dropped , and a sentence from a press release of West Virginia University  with no source.
Anyone have something on this? Otherwise, I'll settle with:
UNKNOWN, data has not been published yet…
Warning: There is currently no independent scientific information on that topic.
The Las Vegas Sands Corporation (a casino and resort company) recently started deploying Oura rings to their staff. They secured 1000 units for their pilot, in the hope it could help them predict COVID-19 before they appear by using measures such as body temperature and heart rate variability.
The $299 Oura ring is a wearable claiming to be "the most accurate sleep and activity tracker". It's designed to monitor sleep, body temperature, heart rate variability and respiratory rate.
On an individual level, users can tag previous days with labels such as cramp, happy, vacation, alcohol, and now cold, flu and covid-19 confirmed, in the hope the app can identify patterns linked to them. The goal is for users to know when an event is "arriving" and when it will be gone. 
In my knowledge, three research groups are working on the topic, so waiting on their results...
(sponsored by Oura)
Study: TemPredict, led by Ashley Mason at UCSF Osher Center for Integrative Medicine 
Goal: data collection of ring + symptoms survey then dev of an algorithm to predict onset of symptoms such as fever, cough and fatigue to identify patterns of onset, progression and recovery for Covid-19
It is ongoing, and has been including front-line healthcare workers and the general population.
No data available yet and they are currently not able to release interim reports.
(partnered with Oura)
Study: "Holistic Integrated Neuro Monitoring" led by Ali Rezai at WVU Rockefeller Neuroscience Institute. 
Goal: early detection of covid-19 symptoms. Dev of an innovative "digital PPE" with an app, AI and stuff
It is ongoing, and has been including front-line healthcare workers
April 8th, press release claiming they can predict symptoms 24 hours prior to onset  (link to the claim, not the source which doesn't exist)
May 28th, press release claiming they can predict the onset of COVID-19 related symptoms (e.g. fevers, coughing, breathing difficulties, fatigue, and others) three days in advance with over 90 percent accuracy.  (same, link to the claim, no source)
It is ongoing, no data to be found
(no link with Oura)
Study led by Jennifer Radin at Scripps Research 
Goal: monitor heart rate and record symptoms to identify areas with viral outbreaks quickly, using different brands and models of wearable
Follow Up to their previous flu-fitbit paper 
 http://webcache.googleusercontent.com/search?q=cache:ryu4PwAzdxsJ:https://wvumedicine.org/rni/covid19/&hl=fr&gl=ch&strip=1&vwsrc=0 (cached because I can't access the page normally)
SARS-CoV-2 binds to.
The style for formatting the references does NOT matter.
We are scientists in various fields, most of us not virologists or epidemiologists, mostly neuroscientists, medical doctors, practitioners, or biologists: see our Facebook page.
Authors: Adrien Chopin, Wing Yan Heidi Wan, Laurent Bichara-Jabour, Kengo Shibata, Carine Beaupere, Jose Carlos Rubio Ballester, Léo Varnet, Caroline Regeasse Girard, Peter Loskill
Special thanks to: Arthur Nguyen, Catherine Agathos, Caroline Conti, Corinna Klinge, Lionel Guilleminot, Cecilia Blikstad, Jean-Baptiste Bernard, Clio Coste, Sandrien van Ommen
Licence: for the moment, we release the current data under the licence CC-by-nc 4.0, so that you are free to share, translate and adapt it, as long you cite us and do not sell the result.