New modeling study published today looks at expected demand for critical care bed space in the U.S.; authors include Seyed Moghadas, @Alison_Galvani, & @JeffDSachs (!). pnas.org/content/pnas/e…
Bottom line: "Our projections indicate that COVID-19 will overwhelm hospital capacity in the United States at the peak of the outbreak."
Authors examined a number of scenarios, focusing on two R0 estimates (2 and 2.5) and also a set of scenarios around cases self-isolating or not at different times after symptoms appear. 
The greatest reduction in peak ICU use happens when symptomatic cases self-isolate within 24 hours of symptoms appearing. The timing of isolation makes more of a difference than the proportion of cases isolating.
For example, "20% of symptomatic individuals with mild illness self-isolating 48 h after symptom onset resulted in comparable outcomes to 5% of them self-isolating 24 h after symptom onset."
"our results indicate that early identification of symptomatic cases combined with timely self-isolation can dramatically reduce the demand for hospital and ICU beds"
The authors are arguing forcefully for rapid isolation of symptomatic cases to save ICU bed space. What we don't want is people out in the world walking around, even with mild symptoms, and infecting others. (social distancing!)
They talk about self-isolation, but this also could apply to symptomatic cases identified through contact tracing. Early identification and isolation of cases is a key intervention for Covid-19, and impacts hospital use.
