New paper suggests #COVID19 cases infected 0.45% of their close contacts. If this seems remarkably small, it’s because it’s likely to be a substantial underestimate of current risk in many areas. Let’s take a closer look... 1/ cdc.gov/mmwr/volumes/6…
This study traced people who’d come into contact with 10 symptomatic early travel-related cases in US. Overall 445 close contacts were identified, and only 2 got infected (i.e. secondary attack rate, SAR = 2/445=0.45%). But this scenario isn’t reflective of an ongoing outbreak 2/
If that contact tracing detected all infections, it implies 10 primary cases generated 2 new infections between them. So the reproduction number R=0.2. But we know in many areas that have outbreaks, R is 2-3 3/. cmmid.github.io/topics/covid19…
In general, R = SAR x contacts. So assuming people have a similar # contacts in ongoing outbreak (44.5 per person), we’d need an SAR among close contacts of at least 5% to get R of 2-3. But 5% is still quite low, so we should also think the situation with closest contacts... 4/
In the US outbreak investigation, 2 out of 19 household members got infected. So the implied SAR here is 11%. We can break transmission down into within and between HH (see image). Average US HH is 2.5 people... 5/ thelancet.com/journals/lance… 
…and to be conservative let’s assume people have 50 contacts outside household. If R=2, it means we have to have to find within and between household secondary attack rates that make this equation work: SARh x 2.5 + SARc x 50 = 2 6/
Clearly SARh=10% and SARc=0.45% doesn’t work in the above equation (R is too small as result). So let’s apply our estimate of 35% (from above Lancet piece) to SARh and (being super pessimistic) multiple the 0.45% value by 5 to get SARc… 7/
These apparently pessimistic assumptions give 35% x 2.5 + 2.3% x 50 = 2. In other words, to get R=2, you need a much higher within and between household transmission than contact tracing investigations in containment phase would suggest. 8/
In the WHO joint mission report, early estimates of household SAR in Guangdong were given as 3-10%. For above reasons, I suspect this is substantial underestimate of risk in populations with ongoing transmission 9/9 who.int/docs/default-s…
Postscript: in above calculation, if household size is 2.5, it of course means that there are 1.5 contacts that can be made, not 2.5 (because the case can't infect themselves). But this doesn't change conclusion – if anything, it only increases the implied transmission risk.
