A tweet⛈️ in the time of #COVID19... (☁️☁️☁️ but with a chance of #optimism ☀️)
There is a LOT of uncertainty / conflicting information, layered onto a LOT of speculation.
So, a roundup of links + charts (with links to sources) and some #SundayThoughts 🤔
There is a LOT of uncertainty / conflicting information, layered onto a LOT of speculation.
So, a roundup of links + charts (with links to sources) and some #SundayThoughts 🤔
2/n
A lot of 🗨️ asking whether people / biz / gov'ts are OVER or UNDER-reacting to #COVID19.
I think this Q drives
- emotional responses (anger, outrage, fear)
- inflammatory politicization (blame-based narratives)
Which I find unhelpful/distracting, bc this:
A lot of 🗨️ asking whether people / biz / gov'ts are OVER or UNDER-reacting to #COVID19.
I think this Q drives
- emotional responses (anger, outrage, fear)
- inflammatory politicization (blame-based narratives)
Which I find unhelpful/distracting, bc this:
3/n
In fluid, rapidly evolving situations w/ incomplete info, vague critiques are just noise.
Better Q:
- WHICH actions are MOST LIKELY to be effective in protecting ourselves and each other, based on best available info re (a) #COVID19 (b) measures being taken in my area?
In fluid, rapidly evolving situations w/ incomplete info, vague critiques are just noise.
Better Q:
- WHICH actions are MOST LIKELY to be effective in protecting ourselves and each other, based on best available info re (a) #COVID19 (b) measures being taken in my area?
4/n
What do (we think) we know re #COVID19 so far?
Generally speaking
- extremely high transmission rates
- relatively low fatality rates
Suggesting
- low individual risk
- high systemic risk
Read on for implications.
Source: McKinsey (3/9/20) tinyurl.com/vunsjed
What do (we think) we know re #COVID19 so far?
Generally speaking
- extremely high transmission rates
- relatively low fatality rates
Suggesting
- low individual risk
- high systemic risk
Read on for implications.
Source: McKinsey (3/9/20) tinyurl.com/vunsjed
5/n
Let's first 👀 at risk of death, not only bc it's the scariest aspect but also bc it's super instructive as to
- how individual risk, though low overall, is UNEVENLY DISTRIBUTED
- how both transmission & fatality rates are HIGHLY CONTEXT-DEPENDENT
Source:
Let's first 👀 at risk of death, not only bc it's the scariest aspect but also bc it's super instructive as to
- how individual risk, though low overall, is UNEVENLY DISTRIBUTED
- how both transmission & fatality rates are HIGHLY CONTEXT-DEPENDENT
Source:
6/n
"GLOBAL" STATS ARE UNHELPFUL bc they combine multiple clusters:
- at differing points in propagation timeline
- with widely varying responses (not only in intensity but in type of action)
Estimates as of today:
Global average: ~3.5%
Lowest (🇰🇷): <1.0%
Highest (🇮🇹): >6.0%
"GLOBAL" STATS ARE UNHELPFUL bc they combine multiple clusters:
- at differing points in propagation timeline
- with widely varying responses (not only in intensity but in type of action)
Estimates as of today:
Global average: ~3.5%
Lowest (🇰🇷): <1.0%
Highest (🇮🇹): >6.0%
7/n
Many factors contribute to ACTUAL differences in fatality rates in 2 major buckets:
- WHO gets it
- HOW effectively they're treated
(Again, bc we are working with incomplete info, we know ⬆️about WHO is at higher risk, ⬇️↔️❓ about HOW to treat... but more on that later)
Many factors contribute to ACTUAL differences in fatality rates in 2 major buckets:
- WHO gets it
- HOW effectively they're treated
(Again, bc we are working with incomplete info, we know ⬆️about WHO is at higher risk, ⬇️↔️❓ about HOW to treat... but more on that later)
8/n
Data from China (useful here bc it gives some depth at 44k cases and sufficient timeline to capture outcomes) indicate older people are at MATERIALLY higher risk, as are those with pre-existing / underlying health conditions.
Data from China (useful here bc it gives some depth at 44k cases and sufficient timeline to capture outcomes) indicate older people are at MATERIALLY higher risk, as are those with pre-existing / underlying health conditions.
9/n
This gives a pretty good clue to help interpret varying figures reported in other clusters.
Italy's underlying demographics explain IN PART the very high fatality rate:
- 23% of residents 65 or older (vs. 15% in US)
- median age 47.3 (vs 38.3 in US)
Source: NYTimes
This gives a pretty good clue to help interpret varying figures reported in other clusters.
Italy's underlying demographics explain IN PART the very high fatality rate:
- 23% of residents 65 or older (vs. 15% in US)
- median age 47.3 (vs 38.3 in US)
Source: NYTimes
10/n
This also reminds us that reliability + utility of reported figures are dependent on other factors, chief among them TESTING.
Limited testing is highly likely to...
- UNDERSTATE scope/pace of transmission
- UNDERSTATE # confirmed cases
- OVERSTATE fatality rates
This also reminds us that reliability + utility of reported figures are dependent on other factors, chief among them TESTING.
Limited testing is highly likely to...
- UNDERSTATE scope/pace of transmission
- UNDERSTATE # confirmed cases
- OVERSTATE fatality rates
11/n
🇰🇷 tested aggressively (KCDC as of 3/15)
- 268k total = 5.3k per million
🇮🇹 tested less (Official #⃣ as of 3/13)
- 86k total = 1.4k per million
⚠️ Stressing these points, by shouting:
- DATA CAN BE MISLEADING WITHOUT PROPER CONTEXT
- DATA COLLECTION MATTERS, A LOT
🇰🇷 tested aggressively (KCDC as of 3/15)
- 268k total = 5.3k per million
🇮🇹 tested less (Official #⃣ as of 3/13)
- 86k total = 1.4k per million
⚠️ Stressing these points, by shouting:
- DATA CAN BE MISLEADING WITHOUT PROPER CONTEXT
- DATA COLLECTION MATTERS, A LOT
12/n
Going back to the Q of "how deadly is #COVID19" I see a LOT of comparisons to the flu. This is ALSO NOT HELPFUL.
We THINK we understand the flu bc it is familiar, but most likely have a poor understanding of "how deadly the flu is" bc WE ARE BAD WITH TINY PROBABILITIES.
Going back to the Q of "how deadly is #COVID19" I see a LOT of comparisons to the flu. This is ALSO NOT HELPFUL.
We THINK we understand the flu bc it is familiar, but most likely have a poor understanding of "how deadly the flu is" bc WE ARE BAD WITH TINY PROBABILITIES.
13/n
So available data says #COVID19 is deadlier than the flu, but that's not very meaningful bc both figures being compared
- are too small for people to grasp easily
- generate skewed multiples that CONFUSE more than CLARIFY
- encourage both OVERreaction and UNDERreaction
So available data says #COVID19 is deadlier than the flu, but that's not very meaningful bc both figures being compared
- are too small for people to grasp easily
- generate skewed multiples that CONFUSE more than CLARIFY
- encourage both OVERreaction and UNDERreaction
14/n
So let's try again. If you're young & relatively healthy, the individual risk to you is low. On that basis, travel restrictions, canceling large events, TP shortage, etc may seem like panic / hysteria.
This is INCORRECT bc many reasons, but mostly bc of SYSTEMIC RISK.
So let's try again. If you're young & relatively healthy, the individual risk to you is low. On that basis, travel restrictions, canceling large events, TP shortage, etc may seem like panic / hysteria.
This is INCORRECT bc many reasons, but mostly bc of SYSTEMIC RISK.
15/n
Going beyond fatality rates...
Time to recovery according to WHO (tinyurl.com/vjrnn3g)
- mild cases ~2 weeks
- severe/critical cases 3-6 weeks
For more useful, experiential comparisons
- mild cases are roughly like the flu
- severe cases are roughly like pneumonia
Going beyond fatality rates...
Time to recovery according to WHO (tinyurl.com/vjrnn3g)
- mild cases ~2 weeks
- severe/critical cases 3-6 weeks
For more useful, experiential comparisons
- mild cases are roughly like the flu
- severe cases are roughly like pneumonia
16/n
Remember from ⬆️ how #COVID19 is HIGHLY contagious?
⚠️Examples of systemic risks
- lots (and lots) of people getting ill
- about 1 in 5 of sick people getting REALLY ill
- lots of people too ill to work for a few weeks
- more ill people than the health system can handle
Remember from ⬆️ how #COVID19 is HIGHLY contagious?
⚠️Examples of systemic risks
- lots (and lots) of people getting ill
- about 1 in 5 of sick people getting REALLY ill
- lots of people too ill to work for a few weeks
- more ill people than the health system can handle
17/n
If/when pandemics reach a tipping point & exceed the capacity of health systems to function effectively, MORE PEOPLE DIE
Those deaths will still be skewed toward the elderly and (sadly) socioeconomically underprivileged. But at least some of these deaths are preventable
If/when pandemics reach a tipping point & exceed the capacity of health systems to function effectively, MORE PEOPLE DIE
Those deaths will still be skewed toward the elderly and (sadly) socioeconomically underprivileged. But at least some of these deaths are preventable
18/n
It is still early days for 🇺🇸 and what we do now matters, A LOT.
Some stuff remains out of individual control (TESTING), but there's still lots that are important and left up to each of us.
It is still early days for 🇺🇸 and what we do now matters, A LOT.
Some stuff remains out of individual control (TESTING), but there's still lots that are important and left up to each of us.
20/n
McKinsey report updated as of today: tinyurl.com/tvg4zgc
"Case Fatality Ratio appears higher where cases are missed and is higher when health systems are overwhelmed"
Yep.
McKinsey report updated as of today: tinyurl.com/tvg4zgc
"Case Fatality Ratio appears higher where cases are missed and is higher when health systems are overwhelmed"
Yep.
