A key feature of response in Japan (mhlw.go.jp/content/109000…) and Korea (imperial.ac.uk/mrc-global-inf…) has been cluster investigation combined with quarantine. New pre-print with @_akiraendo @sbfnk et al looks at dynamics of this approach in more detail: medrxiv.org/content/10.110… 1/ 
If SARS-CoV-2 transmission involves superspreading, then majority of secondary transmission comes from a small proportion of infections. This means that if a new case appears, it's likely to have come from a (potentially quite big) cluster. 2/
Whereas standard 'forward tracing' would be expected to detect up to R secondary infections per case (where R is the reproduction number), this clustering quirk means 'backward tracing' can identify R(1+1/k) additional infections linked to cluster, who can then be traced too. 3/
Under plausible assumptions, we estimated effective backwards + forwards tracing (B) could prevent around twice as many onwards infections as forward tracing alone (A). 4/
In effect, backward tracing is an outbreak investigation process in which new cases and their contacts can be routinely linked via their shared exposure events, supported by real-time data linkage across epidemiological, diagnostic and quarantine datasets. 5/
Ideally, for every new case, it would be possible to answer these questions:
'Were they already identified as contact of a known case?'
'If not, where did this person likely get infected?' 6/
'Were they already identified as contact of a known case?'
'If not, where did this person likely get infected?' 6/
For more on network structure and implications for superspreading and tracing, see this paper: arxiv.org/abs/2005.11283 7/7
