There have been a couple seroprevalence studies pegging the percentage of New Yorkers with SARS-CoV-2 antibodies at 15 to 20% or more, but this is the first I've seen with time-phasing. What struck me most is that at the end of March, seroprevalence in the general population was around two percent. Two percent doesn't seem like a lot, but 2% of a place the size of New York City means over 100,000 people.
If you consider at least a 2-week lag between infection and antibody presence, then by mid-March over 100,000 cases were walking around New York. Confirmed cases for NY in mid-March numbered fewer than 1000. So the left part of the NY curve is capturing maybe one of every 100-200 cases.
During the first two weeks of March, NY had likely been seeded with thousands of infected people, with zero social distancing measures in place. Eventually, well over 1M New Yorkers would be infected (based on seroprevalence studies).
FL, TX, AZ and CA probably aren't capturing anywhere close to 100% of cases in official case counts, but it's certainly closer to 1 in 10 than 1 in 100. So even if the left part of the curve in those places looks like the left part of NY's curve, the situations they depict are very different.
I noticed the drop in ages in Florida a couple of weeks ago when I was spot-checking counties that were at the leading edge of increases in the state. I was pretty astonished to see the aggregate median age in Palm Beach County drop from 53 to 45 between May 10 and June 15 given how many cases PBC already had on the books at the start of that period.
A few days later, it became more prominent because daily median ages were added to the county reports (daily PDF that gives a snapshot for each county in the state). Going through those last night, I noticed some wicked low median ages occasionally in the rural counties (and it wasn't just instances of the median age being 16 because the only person in the county who tested positive that day was 16). Reading your post reminded me that I could look at the line data and see what comprised those medians. In those instances, there were a few infants/toddlers in the mix.
They don't amount to a lot, of course, but those are age groups where the number of cases have more than doubled in the past two weeks. Going back a bit further, to June 1, there were 1,014 people in the 5-14 age bracket compared to 3,002 in the 85+ group. As of yesterday the younger group numbered 4,297 compared to 4,742 in that oldest group (And looking at the data Liora compiled, I notice that there's a crossover of more cases being in the youngest group than the oldest group in LA County.) Obviously doesn't cause the same impact to the median age as the 20 and 30 somethings, but the youngest people are pulling down that median as well.
There was a local news story today that kind of ties into that (as well as your point that we're seeing data now that we didn't see in the initial run-up of cases)
"Wolfson Children’s Hospital said its emergency room is seeing an increase in children being tested for COVID-19.
The hospital told News4Jax about the rise in pediatric testing after the newsroom received an emailed news tip from a man who claimed he heard the hospital was at full capacity with children suffering from multiple organ failures as a result of COVID-19. According to the claim, the hospital staff was overwhelmed by these pediatric cases and no one in news media was talking about it.
In response, Wolfson released the following statement:
'Wolfson Children’s Hospital is not full. Our census is under 80% at this time. Patients are hospitalized for a variety of conditions, from prematurity-related issues in our NICU to post-surgical care, along with treatment for childhood cancer and congenital heart defects.
We have seen an uptick in children being tested at our Wolfson Children’s ERs. While our ERs are available for children’s emergencies, and we urge people not to delay care for emergency situations, we want to make sure parents and guardians are aware there are pediatric COVID-19 testing sites throughout the community...'"
The thing I like about Substack is that there is an ability to limit comments to paid subscribers. For me, this isn't even about the money, it's about the comments. I love the ability to filter out the nonsense so that I only hear from people who actually care about engaging with the ideas here and making kind and thoughtful arguments. Glad you're here! :)
Re the age shift in cases and Arizona. Here in Maricopa County (60% of AZ population) the situation has changed greatly since the start of the surge on May 25. Up till then those under 45 were 47% of cases, since then 71%. And, so far, hospitalization and mortality rates for all ages have significantly declined. The problem is the sheer number of cases is placing stress on the system. As of June 29 the case count is almost 8X that of May 25, and while it's increased much faster in the younger brackets the count for 65+ is still up 3X. And while deaths have only increased slightly the ratios have changed, prior to May 25, 12% of deaths were under 65, since then 25%.
Fellow Maricopa County resident speaking - One of the things the change in the death demographics over time strongly suggests (assuming that relative IFR by age has remained constant) is that there has been an age shift in the actual infections (not just a shift in the demographics being tested, and hence the official case count). This presumably indicates that we're seeing some level of success in preferentially protecting vulnerable age demographics.
As a semi-quantitative follow-up, if we (1) calculate first order CFR's by age by dividing total deaths as of June 9 by total cases as of May 25 (assuming two-week lag from case ID to death) and (2) assume the CFR by age for the current surge will be unchanged - I hope and at least semi-reasonably think it will be lower (though by how much I don't want to guess, lest I be proved embarrassingly wrong) - that suggests deaths from the current surge will be only one-third as much as if we projected based on the age demographics from the first wave. I will show my work if absolutely forced to.
Some more on the contrast between % and numbers. On the May 25 baseline, 6% of cases between 20-44 required hospitalization. That is now down to 2% but the number hospitalized has tripled. 15% of those between 45-64 required hospitalization, that's now down to 7%, but the number has doubled, which is why the hospitals are now stressed. Those in Long Term Care (Maricopa tracks these separately) have done reasonably well as cases have slightly less than doubled but those over 65+ and NOT in LTC have increased by 5.5X, even as they've shrunk as a % of overall cases in the county.
Some news from Minnesota, today we saw the first significant bump in hospitalizations and deaths (we had been 10 or lower daily deaths) for the first time in about a week and a half. I wonder if this might be due to the coming holiday over the next few days, perhaps the reporting flooded in today and we'll see drops over this weekend. The real trend will appear on Tuesday and Wednesday next week. Here's our data from Minnesota for those who want to take a look: https://www.health.state.mn.us/diseases/coronavirus/situation.html
Re: the "fog or war" in the early numbers of the pandemic:
Here's a preprint of a longitudinal seroprevalence study for NYC: https://www.medrxiv.org/content/10.1101/2020.06.28.20142190v1.full.pdf
There have been a couple seroprevalence studies pegging the percentage of New Yorkers with SARS-CoV-2 antibodies at 15 to 20% or more, but this is the first I've seen with time-phasing. What struck me most is that at the end of March, seroprevalence in the general population was around two percent. Two percent doesn't seem like a lot, but 2% of a place the size of New York City means over 100,000 people.
If you consider at least a 2-week lag between infection and antibody presence, then by mid-March over 100,000 cases were walking around New York. Confirmed cases for NY in mid-March numbered fewer than 1000. So the left part of the NY curve is capturing maybe one of every 100-200 cases.
During the first two weeks of March, NY had likely been seeded with thousands of infected people, with zero social distancing measures in place. Eventually, well over 1M New Yorkers would be infected (based on seroprevalence studies).
FL, TX, AZ and CA probably aren't capturing anywhere close to 100% of cases in official case counts, but it's certainly closer to 1 in 10 than 1 in 100. So even if the left part of the curve in those places looks like the left part of NY's curve, the situations they depict are very different.
I noticed the drop in ages in Florida a couple of weeks ago when I was spot-checking counties that were at the leading edge of increases in the state. I was pretty astonished to see the aggregate median age in Palm Beach County drop from 53 to 45 between May 10 and June 15 given how many cases PBC already had on the books at the start of that period.
A few days later, it became more prominent because daily median ages were added to the county reports (daily PDF that gives a snapshot for each county in the state). Going through those last night, I noticed some wicked low median ages occasionally in the rural counties (and it wasn't just instances of the median age being 16 because the only person in the county who tested positive that day was 16). Reading your post reminded me that I could look at the line data and see what comprised those medians. In those instances, there were a few infants/toddlers in the mix.
They don't amount to a lot, of course, but those are age groups where the number of cases have more than doubled in the past two weeks. Going back a bit further, to June 1, there were 1,014 people in the 5-14 age bracket compared to 3,002 in the 85+ group. As of yesterday the younger group numbered 4,297 compared to 4,742 in that oldest group (And looking at the data Liora compiled, I notice that there's a crossover of more cases being in the youngest group than the oldest group in LA County.) Obviously doesn't cause the same impact to the median age as the 20 and 30 somethings, but the youngest people are pulling down that median as well.
There was a local news story today that kind of ties into that (as well as your point that we're seeing data now that we didn't see in the initial run-up of cases)
"Wolfson Children’s Hospital said its emergency room is seeing an increase in children being tested for COVID-19.
The hospital told News4Jax about the rise in pediatric testing after the newsroom received an emailed news tip from a man who claimed he heard the hospital was at full capacity with children suffering from multiple organ failures as a result of COVID-19. According to the claim, the hospital staff was overwhelmed by these pediatric cases and no one in news media was talking about it.
In response, Wolfson released the following statement:
'Wolfson Children’s Hospital is not full. Our census is under 80% at this time. Patients are hospitalized for a variety of conditions, from prematurity-related issues in our NICU to post-surgical care, along with treatment for childhood cancer and congenital heart defects.
We have seen an uptick in children being tested at our Wolfson Children’s ERs. While our ERs are available for children’s emergencies, and we urge people not to delay care for emergency situations, we want to make sure parents and guardians are aware there are pediatric COVID-19 testing sites throughout the community...'"
(https://www.news4jax.com/news/local/2020/06/30/wolfson-er-sees-uptick-in-children-being-tested-for-covid-19/)
This is some great digging! Mind if I point to this comment in the next newsletter?
Do as you please, please!
Kind of poignant but her 'citizen journalism' sounds very much like the blog era circa 2002-2012 (?)
Twitter can be something similar, as can Substack, evidently.
The thing I like about Substack is that there is an ability to limit comments to paid subscribers. For me, this isn't even about the money, it's about the comments. I love the ability to filter out the nonsense so that I only hear from people who actually care about engaging with the ideas here and making kind and thoughtful arguments. Glad you're here! :)
Re the age shift in cases and Arizona. Here in Maricopa County (60% of AZ population) the situation has changed greatly since the start of the surge on May 25. Up till then those under 45 were 47% of cases, since then 71%. And, so far, hospitalization and mortality rates for all ages have significantly declined. The problem is the sheer number of cases is placing stress on the system. As of June 29 the case count is almost 8X that of May 25, and while it's increased much faster in the younger brackets the count for 65+ is still up 3X. And while deaths have only increased slightly the ratios have changed, prior to May 25, 12% of deaths were under 65, since then 25%.
Fellow Maricopa County resident speaking - One of the things the change in the death demographics over time strongly suggests (assuming that relative IFR by age has remained constant) is that there has been an age shift in the actual infections (not just a shift in the demographics being tested, and hence the official case count). This presumably indicates that we're seeing some level of success in preferentially protecting vulnerable age demographics.
As a semi-quantitative follow-up, if we (1) calculate first order CFR's by age by dividing total deaths as of June 9 by total cases as of May 25 (assuming two-week lag from case ID to death) and (2) assume the CFR by age for the current surge will be unchanged - I hope and at least semi-reasonably think it will be lower (though by how much I don't want to guess, lest I be proved embarrassingly wrong) - that suggests deaths from the current surge will be only one-third as much as if we projected based on the age demographics from the first wave. I will show my work if absolutely forced to.
Some more on the contrast between % and numbers. On the May 25 baseline, 6% of cases between 20-44 required hospitalization. That is now down to 2% but the number hospitalized has tripled. 15% of those between 45-64 required hospitalization, that's now down to 7%, but the number has doubled, which is why the hospitals are now stressed. Those in Long Term Care (Maricopa tracks these separately) have done reasonably well as cases have slightly less than doubled but those over 65+ and NOT in LTC have increased by 5.5X, even as they've shrunk as a % of overall cases in the county.
Some news from Minnesota, today we saw the first significant bump in hospitalizations and deaths (we had been 10 or lower daily deaths) for the first time in about a week and a half. I wonder if this might be due to the coming holiday over the next few days, perhaps the reporting flooded in today and we'll see drops over this weekend. The real trend will appear on Tuesday and Wednesday next week. Here's our data from Minnesota for those who want to take a look: https://www.health.state.mn.us/diseases/coronavirus/situation.html