Read on Twitter
1/ Fascinating new internal hospital report from Houston coming soon; the five-word takeaway sums it up.
Meantime, @TheAtlantic has some hot panic porn suggesting deaths in the Sunbelt could reach New York City levels per-capita - 550,000 deaths! That is... unlikely. Why?
Meantime, @TheAtlantic has some hot panic porn suggesting deaths in the Sunbelt could reach New York City levels per-capita - 550,000 deaths! That is... unlikely. Why?
2/ A whole bunch of reasons. Three you may know, the fourth and fifth are more subtle.
1: CASE MIX. The age of the infected is by far the most important factor in deaths, and reported infections have skewed much younger in the Sunbelt. Even @theatlantic admits this.
1: CASE MIX. The age of the infected is by far the most important factor in deaths, and reported infections have skewed much younger in the Sunbelt. Even @theatlantic admits this.
Cont/
2: Better treatments - remdesivir, steroids, proning - are reducing death rates.
3: NYC hospitals overused ventilators and killed patients; that's not happening now. And NYC hospitals were extremely stressed in late March/early April, leading to subpar care at times.
2: Better treatments - remdesivir, steroids, proning - are reducing death rates.
3: NYC hospitals overused ventilators and killed patients; that's not happening now. And NYC hospitals were extremely stressed in late March/early April, leading to subpar care at times.
Cont/
The more subtle factors.
4: For reasons we don't understand, death rates relative to reported cases seem to vary widely by state and country. What's odd is they get established early and then don't change much. Example: Germany has a much lower rate than Italy. Why?
The more subtle factors.
4: For reasons we don't understand, death rates relative to reported cases seem to vary widely by state and country. What's odd is they get established early and then don't change much. Example: Germany has a much lower rate than Italy. Why?
Cont/
Lots of potential explanations; maybe Germany tested more widely. Maybe Germans are healthier. Maybe there's some subtle genetic protection (unlikely). Maybe care in Germany is better. The point is - for whatever reason, these differences DO NOT SEEM TO FADE.
Lots of potential explanations; maybe Germany tested more widely. Maybe Germans are healthier. Maybe there's some subtle genetic protection (unlikely). Maybe care in Germany is better. The point is - for whatever reason, these differences DO NOT SEEM TO FADE.
Cont/
And the Sunbelt states are playing the role of Germany to New York's Italy right now. Further, deaths seem to follow the same pattern everywhere - they top out quickly and then plateau for a while. New York got crushed not because it had 600 deaths one day...
And the Sunbelt states are playing the role of Germany to New York's Italy right now. Further, deaths seem to follow the same pattern everywhere - they top out quickly and then plateau for a while. New York got crushed not because it had 600 deaths one day...
Cont/
But because it had 600 deaths a day for weeks. We don't know where Sunbelt deaths will top out yet - but we know they are rising much more slowly than they did in New York BOTH in absolute terms and relative to hospitalizations, which suggests the peak will be much lower.
But because it had 600 deaths a day for weeks. We don't know where Sunbelt deaths will top out yet - but we know they are rising much more slowly than they did in New York BOTH in absolute terms and relative to hospitalizations, which suggests the peak will be much lower.
Cont/
Finally, point 5: @theatlantic overstates time lag from symptom onset (or a positive test) to death. @cdcgov reports an average of a 15-day delay from symptoms to death. Infection to death is reported around 23 days on average; but testing takes place days after infection.
Finally, point 5: @theatlantic overstates time lag from symptom onset (or a positive test) to death. @cdcgov reports an average of a 15-day delay from symptoms to death. Infection to death is reported around 23 days on average; but testing takes place days after infection.
In New York City, in fact, the lag was even shorter - possibly because cases went uncaught early, but possibly because the case mix was so bad and possibly because of problems with care. Why does this matter?
The Sunbelt spike in positive tests is now five weeks-plus old. Even with better care, even with a better case mix, the sharp spike in deaths should be happening now - if it's going to. But in reality, there's increasing evidence that hospitalizations have peaked in TX and AZ...
Which is a very positive sign, because hospitalizations LEAD; deaths lag. We can't be sure for another week or two, but the weight of the evidence overall STRONGLY suggests that the notion that the Sunbelt states will approach New York levels of mortality is way off.
