Thread by @AnnaRotkirch

New country data & simulations show why
EXTENT & TIMING of #masks4all matter in Covid-19 exits:
Universal Masking to Restart Society and Save Lives bit.ly/2KxxjrW
Summing arguments here, 1/n
thanks co-authors @AlekseyMorgunov @dekai123 @guypgoldstein @V_alien

2/n We made two different models to predict what we gain from universal face mask wearing.
The simulation model by @AlekseyMorgunov uses a stochastic dynamic network SEIR (susceptible-exposed-infectious-recovered) approach.

3/n In the SEIR model, nodes=individuals, edges= interactions in a network. Each node has a state (S, E, I, R, or F (dead)).
Adjacent nodes=close contacts, contacts from anywhere =global contacts.
Initial infected set to 1%.

4/n SEIR model shows impact of masking with social distancing, compared to lockdown, and to distancing only.
Timeline as UK+many Western countries:
lockdown March 24th (day 1), lifted on May 31st.
Universal masking from April.
Simulation 500 days from day 0, appr 17 months.

5/n The simulation indicates that population where 80% use masks (red) +socially distance flattens curve EVEN MORE than strict lock-down (blue).
Figure by
@AlexeyMorgunov

6/n SEIR model also estimates lives saved:
Masking at 80% +distancing (red) => 60,000 deaths
Masking at 50% +distancing (orange) => 240,000 deaths
Strict lockdown (blue) => 180,000 deaths (and probably not socially feasible), by
@AlekseyMorgunov

7/n The second model is a stochastic individual agent simulation of universal masking effects by
@dekai123 and is neatly explained in

8/n This model has an ABM approach, which allows us to put masks on ’individuals’ & assign properties to masks.
It shows why face masks — even non-medical cloth masks — could be surprisingly effective.

9/n New country data support simulation results from above:
Universal Masking is associated with lower daily Covid-19 daily growth, faster peak reduction.
(and maybe no lockdowns at all)

10/n Data on mask wearing is often absent from corona policy comparisons, so @guypgoldstein
and @AlexeyMorgunov compiled new data.
Countries / regions who best manage Covid-19 have
(i) established universal masking cultures, or
(ii) mandatory or recommended universal masking.

11/n Countries or regions performing this well include Taiwan, South Korea, Singapore, Japan, Hong Kong, Macau, and Beijing, Shanghai, Guang-dong.

12/n
Perhaps even more crucially:
countries which do test, track, and quarantine, but lack universal masking, did less well in COVID-19 epidemic control as of April 10, 2020.

13/n This new country comparison is fully in line with the substantial effects of masking reported e.g. by
@Jabaluck @NAChristakis
in papers.ssrn.com/sol3/papers.cf…

The take homes:
Masking should be universal to work
=all who can safely do so in a population.
50% (not enough) is different from 80% (good).
#masks4all

Finally, the TIMING of universal masking matters.
Mass masking introduced now can affect timing and shape of possible autumn Covid-19 peaks.
#Masks4All

ping @aivelo @Virusvinkki @vesa_vihriala @THLorg @MarinSanna @niinisto @AulaMK @MariaOhisalo @halkka @KirsiVarhila @JariSaramaki @jeremyphoward @mariholopainen @CSalmivalli

my co-author @AlexeyMorgunov keeps getting wrong tags from me, but I assure he is real, brilliant, and looks like a lama on Twitter

this is the real @AlexeyMorgunov who made those models. Complicated even for Twitter bots to tell them apart clearly... apologies

Here is the true Twitter account of @AlexeyMorgunov (there are 2). If it looks like a lama but models like a god, it's him. As in: researchgate.net/publication/34…