Thread by @mlipsitch

In places with many known & suspected COVID cases it is getting really intense. Efforts to do contact tracing are overwhelming even the top health departments nytimes.com/2020/03/09/nyr…. At this stage it is wrongheaded to pour exponentially growing resources into this strategy.

It is unsustainable (effort grows with epidemic size). We need measures that while painful for all will slow social contact - cancelling public gatherings, paid sick leave, working from home, and the like. Social distancing is the general name for these interventions.

And we need to stop feeling sheepish about it and just realize that some places (Italy, Iran) are in crisis, and some are very likely in the days before crisis, a crisis that will be less bad if we slow down the virus. #flattenthecurve to reduce peak demand on health care

The 1918 analogies have been discussed a lot and are correect. Flattening the curve reduces health care load, delays risk for everyone, and reduces total epidemic size. Here are some data from a paper we submitted to @medrxivpreprint and is awaiting clearance. Led by @ruoranepi

with @megan_b_murray @cmyeaton Eric Toner, Qi Tan. The link to full document will be on last tweet as I want to replace it with a more permanent one.

We looked at the epidemics in Wuhan and Guangzho and their ICU and hospital bed use vs US capacity. Summary: Wuhan's peak critical case load per capita was equal to the total number of ICU beds per capita in the US -- a similar experience to Wuhan would fill our ICU with COVID

Here's the graph. Gray lines show US capacities for ICU and hospital beds

Guangzho intervened much earlier in their epidemic, and had a MUCH smaller peak demand. Note different scale

My takeaways: 1) early intervention spares the health system from intense stress -- like Philly vs. St. Louis.
2) Early intervention means before it feels bad. Guangzho intervened when they had 7 confirmed cases & 0 deaths. Wuhan's came when they had 495 confirmed cases, 23 dead

3) We will not intervene as intensely as China, making speed even nore important.
4) Slowing transmission did not immediately relieve health care burden. People take a long time (weeks) to get really sick, so the peak burden trailed peak transmission by weeks esp in Wuhan

Here is a link to the preprint on my dropbox. Will replace with @MedArXiv or Harvard DASH when these clear dropbox.com/s/e7hj97y73ym2…

@threadreaderapp unroll

Here is a permanent link nrs.harvard.edu/urn-3:HUL.Inst…

Now on DASH, Harvard's excellent repository dash.harvard.edu/handle/1/42599…

And here is a much more readable figure from @ruoranepi - same data plotted more clearly