Thread by @CJinKnoxTN

A thread about ICU Bed/Ventilator shortages and actual data for COVID-19 are telling us about expert model assumptions. /1

Gov Cuomo of NY stated on Saturday, 3/28/2020, "New York has tested 155,000 people...Projections show New York will reach its apex of cases in 14 to 21 days." So far, so good. /2
syracuse.com/coronavirus/20…

But then he went on to say "The state would need 140,000 hospital beds at that time, including 40,000 ICU beds with ventilators. The state is still in need of 30,000 ventilators." /3

This indicates he has 10,000 ventilators now, about 4,300 were already in NY hospitals (according to article below), and the remaining 5700 ventilators he must have sourced since outbreak (4000 from Fed stockpile?). /4
thecity.nyc/2020/03/new-yo…

In the Saturday briefing, Cuomo reported the state currently has 1755 patients in ICU beds, 7328 hospitalizations, and 52,318 cases. So we have data from field: NY hospital admission rate is 14% of confirmed cases (7328/52318) and ICU admission rate is 3.3% (1755/52318). /5

He says he needs 30K more ventilators for his apex in 2-3 weeks. That effectively means he needs 1,192,433 confirmed cases over the next 2-3 wks! (40,000/.033). NY State confirmed daily cases are averaging around 7K. /6

From 7K/day, NY will need to grow confirmed new cases/day by 15% of cumulative total each day to reach 1.2M. Since positive rate is 40% (at most) in NYC (much lower elsewhere), it will need to administer about 3 million tests over the next 14-21 days!! (1.2M/0.4) /7

Anything is possible, of course. But that's a very heavy lift. More likely is the model assumptions he is working with haven't been updated quickly enough with real data from field. /8

I found granular data created by the Harvard Global Health Institute (HGHI). So now I can see US national, but also state-by-state, and down to each of the 305 US Hospital Referral Regions (HRR). HGHI data was the basis for several news articles (maybe many more). /9

In the data provided at bottom of linked article below, the model HGHI uses is baked into the data. But it shows that from best case scenario to worst, HGHI effectively assumes a 20.8% hospital admission rate and a 4.5% ICU admission rate. /10
globalepidemics.org/2020-03-17-car…

Using HGHI worst case scenario (60% infection of population over 6 months!), NY only needs 17,000 more ventilators - not 30,000 more. But in worst case scenario in miniature (cruise ship with shared ventilation in enclosed space) you didn't see 60% penetration of infection /11

Using HGHI middle road (40% infection over 6 months) projection, NY only needs 7900 more ventilators. Using HGHI low range of 20% infection of population over 6 months, NY will have a surplus of about 1000 ventilators. /12

Using HGHI model but adjusted to field data rates for ICU adm rate of 3.3% (40% infection over 6 months - still very aggressive) projection, NY only needs 3300 more ventilators. /13

Using HGHI model but adjusted to NY field data rates for ICU adm rate of 3.3% (20% infection over 6 months) projection, NY has a surplus of 3349 ventilators. /14

As indicated by the tables, I've adapted model to modify key assumptions if data changes from the field. IOW, if the mean ICU stay could be cut from 12 days to 6 due to new therapies, things look much better. ;) /15

Anyway, just hope the data lends better perspective. We all are hoping and praying that even the best case scenarios improve. /16 END