Evidence that social distancing may be working

Evidence that social distancing may be working - Printable Version

+- The CardBoard (https://thecardboard.org/board)
+-- Forum: Emergency (https://thecardboard.org/board/forum-11.html)
+--- Forum: Covid-19 (https://thecardboard.org/board/forum-12.html)
+--- Thread: Evidence that social distancing may be working (/thread-19224.html)

Pages: 1 2

Evidence that social distancing may be working - BostonCard - 03-25-2020

[tweet]https://twitter.com/TopherSpiro/status/1242601771131973633?s=20[/tweet]

The "fever maps" and analyses that Kinsa (a connected thermometer) have been running are a pretty cool demonstration proof of concept. I am a touch weary of the analysis, but it seems to comport with reality, and if the data proves reliable, it might be able to give decision-makers an idea of where people are starting to get fevers a couple of days before they start showing up to ERs and testing positive for coronavirus. Anyways, again, if the data is showing us what we think it shows us*, it seems to suggest that social distancing is working.

BC

* I am a bit skeptical of the data displayed in the plots above because given what we know about the disease you would expect more lag time between when a social distancing measure is taken and a reduction in new cases; the plot above suggests it is a very short lag.

RE: Evidence that social distancing may be working - cardcrimson - 03-25-2020

The drop off was much slower as expected in Santa Clara. Miami is indeed odd. It is, and has been, in the 80s with humidity, don't know if that helps.

RE: Evidence that social distancing may be working - burger - 03-25-2020

Is there any evidence that this thermometer data actually measures sickness levels at the population level? How many of these are out there? I'm skeptical that lockdowns can have such quick effects.

RE: Evidence that social distancing may be working - oldalum - 03-25-2020

I too think the time from implementation of mandatory measures to decrease in excess numbers of fevers is too short to be causal, but one explanation may be some voluntary adoption of social distancing before the date of a mandatory measure. My family, in Santa Clara County, was doing it days before it was required (we may be atypical, I don't know)

RE: Evidence that social distancing may be working - 82 Card - 03-25-2020

Miami may have shipped its cases to other states as soon as the measures started. Not much point in hanging around if the beach is closed.

RE: Evidence that social distancing may be working - BostonCard - 03-25-2020

(03-25-2020, 01:30 PM)82 Card Wrote: Miami may have shipped its cases to other states as soon as the measures started. Not much point in hanging around if the beach is closed.

How many people bring their connected thermometer with them when they go to the beach?

BC

RE: Evidence that social distancing may be working - stupac2 - 03-25-2020

Remember that this picks up all fevers, it's not just covid fevers. The decline would happen more quickly because of fevers from pathogens with shorter incubation times, which is basically every common illness, right? It's not like SCC was running at this stable background level of covid fevers, it was flu, bad colds, etc etc. Those are what's coming down, covid is barely measurable.

They have some maps that show "anomalous" fevers that, provided whatever algorithm they use to predict the number of fevers can be trusted, would be much better at spotting covid in a population that's otherwise not changing its behaviors. But I'm not sure you can safely pick out a few dozen cases of covid out of what must be tens of thousands of background fevers.

RE: Evidence that social distancing may be working - 82 Card - 03-25-2020

(03-25-2020, 01:50 PM)BostonCard Wrote:
(03-25-2020, 01:30 PM)82 Card Wrote: Miami may have shipped its cases to other states as soon as the measures started. Not much point in hanging around if the beach is closed.

How many people bring their connected thermometer with them when they go to the beach?

BC

College students aren't the only ones that go to Florida this time of year. I don't know anyone who has one of these thermometers. Seems like the kind of thing well-off parents would use on their kids or give to granny so they keep an eye on them.

RE: Evidence that social distancing may be working - oldalum - 03-25-2020

(03-25-2020, 01:59 PM)stupac2 Wrote: The decline would happen more quickly because of fevers from pathogens with shorter incubation times, which is basically every common illness, right? Yes, good point!

But I'm not sure you can safely pick out a few dozen cases of covid out of what must be tens of thousands of background fevers.
I assume they are applying appropriate sophisticated statistical techniques to determine what is a significant signal. Their consultants and peer-reviewed publication evidence this sophistication in the past. They have not yet published this study with its methodology.

RE: Evidence that social distancing may be working - M T - 03-25-2020

(03-25-2020, 01:59 PM)stupac2 Wrote: Remember that this picks up all fevers, it's not just covid fevers. The decline would happen more quickly because of fevers from pathogens with shorter incubation times, which is basically every common illness, right? It's not like SCC was running at this stable background level of covid fevers, it was flu, bad colds, etc etc. Those are what's coming down, covid is barely measurable.

They have some maps that show "anomalous" fevers that, provided whatever algorithm they use to predict the number of fevers can be trusted, would be much better at spotting covid in a population that's otherwise not changing its behaviors. But I'm not sure you can safely pick out a few dozen cases of covid out of what must be tens of thousands of background fevers.

Even so, this data suggests that people are reducing the spread of disease. One could reasonably expect that a reduction of temperatures, indicating a reduction for shorter incubation diseases would also indicate that the spread of COVID-19 is being impacted. This may be a good measure of effectiveness of social distancing efforts (closing schools, limiting gatherings, etc) in impacting disease spread. Certainly better than miles driven (gasoline use, etc.). Suburban electrical usage may be a way to measure time that homes are occupied.

The tweet's data is a bit out of date. For Santa Clara County, the Kinsa estimate of flu-like illnesses is now at 1.94%, just off the bottom of that chart. It appears to be starting to level out.

RE: Evidence that social distancing may be working - akiddoc - 03-25-2020

We were seeing a ton of Influenza until early March. It seems to have come to an abrupt halt. The numbers are more likely reflective of that since it is (was) much more prevalent and causes a higher fever. It does show that social isolation can stop the spread of viral illness, though.

RE: Evidence that social distancing may be working - BostonCard - 03-25-2020

(03-25-2020, 03:43 PM)akiddoc Wrote: We were seeing a ton of Influenza until early March. It seems to have come to an abrupt halt. The numbers are more likely reflective of that since it is (was) much more prevalent and causes a higher fever. It does show that social isolation can stop the spread of viral illness, though.

Yes, Hong Kong reported this last month. Basically, the rates of positives of all diseases, not even just viral respiratory illnesses, plummeted. Wish I still had the plot.

BC

RE: Evidence that social distancing may be working - stupac2 - 03-25-2020

(03-25-2020, 03:10 PM)oldalum Wrote:
(03-25-2020, 01:59 PM)stupac2 Wrote: The decline would happen more quickly because of fevers from pathogens with shorter incubation times, which is basically every common illness, right? Yes, good point!

But I'm not sure you can safely pick out a few dozen cases of covid out of what must be tens of thousands of background fevers.
I assume they are applying appropriate sophisticated statistical techniques to determine what is a significant signal. Their consultants and peer-reviewed publication evidence this sophistication in the past. They have not yet published this study with its methodology.

I'm sure it's sophisticated, but this is just a really hard problem. You have a small (but probably increasing) signal riding on a large baseline and now that baseline is changing rapidly because of an intervention that's trying to affect the signal. I'm not sure how you predict what's supposed to happen there, it's a unique situation so the previous years' data won't be too helpful.

RE: Evidence that social distancing may be working - burger - 03-25-2020

(03-25-2020, 04:42 PM)stupac2 Wrote:
(03-25-2020, 03:10 PM)oldalum Wrote:
(03-25-2020, 01:59 PM)stupac2 Wrote: The decline would happen more quickly because of fevers from pathogens with shorter incubation times, which is basically every common illness, right? Yes, good point!

But I'm not sure you can safely pick out a few dozen cases of covid out of what must be tens of thousands of background fevers.
I assume they are applying appropriate sophisticated statistical techniques to determine what is a significant signal. Their consultants and peer-reviewed publication evidence this sophistication in the past. They have not yet published this study with its methodology.

I'm sure it's sophisticated, but this is just a really hard problem. You have a small (but probably increasing) signal riding on a large baseline and now that baseline is changing rapidly because of an intervention that's trying to affect the signal. I'm not sure how you predict what's supposed to happen there, it's a unique situation so the previous years' data won't be too helpful.

Here's a paper using this data (or a similar dataset): https://academic.oup.com/cid/article/67/3/388/4838992

It's actually a fairly large sample size. I'm still not sure how useful this is on a county-by-county level, as in those maps. In the paper, they use 10 CDC regions in the US. Sample sizes by county must be a couple of orders of magnitude smaller.

RE: Evidence that social distancing may be working - stupac2 - 03-25-2020

I'm a bit too on beer right now to read an academic paper, but my issue was never with their sample size, just that finding a signal, any signal, that's multiple orders of magnitude lower than your baseline is always a very hard problem. Then you add the confounding factor of the massive behavioral changes being implemented now and I'm not sure it's possible at all.

RE: Evidence that social distancing may be working - cardcrimson - 03-25-2020

(03-25-2020, 05:59 PM)stupac2 Wrote: I'm a bit too on beer right now to read an academic paper, but my issue was never with their sample size, just that finding a signal, any signal, that's multiple orders of magnitude lower than your baseline is always a very hard problem. Then you add the confounding factor of the massive behavioral changes being implemented now and I'm not sure it's possible at all.

Damn it. I just spit my wine onto my keyboard. . . .Thanks for the laugh!

RE: Evidence that social distancing may be working - BostonCard - 03-25-2020

Alternative explanation now that I think about it:

Enact a shelter-in-place order for a deadly pandemic and all of a sudden a lot of hypocondriacs (you know this because they bought a connected thermometer [ed: I bought one too, lol]) and a lot of hypocondriacs are stuck at home with nothing better to do than check their temperature again, driving down the proportion of abnormal temperatures out there.

BC

RE: Evidence that social distancing may be working - stupac2 - 03-25-2020

(03-25-2020, 08:24 PM)cardcrimson Wrote:
(03-25-2020, 05:59 PM)stupac2 Wrote: I'm a bit too on beer right now to read an academic paper, but my issue was never with their sample size, just that finding a signal, any signal, that's multiple orders of magnitude lower than your baseline is always a very hard problem. Then you add the confounding factor of the massive behavioral changes being implemented now and I'm not sure it's possible at all.
Damn it. I just spit my wine onto my keyboard. . . .Thanks for the laugh!

Glad I could help!

(03-25-2020, 08:31 PM)BostonCard Wrote: Alternative explanation now that I think about it:

Enact a shelter-in-place order for a deadly pandemic and all of a sudden a lot of hypocondriacs (you know this because they bought a connected thermometer [ed: I bought one too, lol]) and a lot of hypocondriacs are stuck at home with nothing better to do than check their temperature again, driving down the proportion of abnormal temperatures out there.

BC

Amusingly this came up in a different venue where I posted this website, but these charts are just showing the total number of fevers, so this explanation doesn't make sense (since it's not per reading or per capita or anything else, although presumably it adjusts for the total number of thermometers). The only way this explanation would work is if you think that a bunch of people with low-grade fevers who otherwise wouldn't check it are suddenly checking, and that would drive it in the opposite direction (and, as you said, the people who buy fancy-ass thermometers are probably the kind to check out of caution every time they feel remotely wrong, so it doesn't even really make sense then).

RE: Evidence that social distancing may be working - BostonCard - 03-25-2020

As best I can tell, the y-axis is %-ill. The question is what is the denominator of that percentage: all people who have a connected thermometer in the county, or the % of those who tested who are abnormal? What is the numerator? Number of total tests, or number of thermometers (really users, because [sheepishly admitting I have one*] the app lets you pick whose temperature you took.

BC

* In my defense, I had no idea there was an app for it when I bought the thermometer; it was just a fairly user-friendly with good reviews on Amazon.

RE: Evidence that social distancing may be working - stupac2 - 03-25-2020

I forgot that you started this with that tweet, this is the data I was thinking of (same company, different format): https://healthweather.us/

I think this is a useful tool, but that you have to be very careful parsing the data because so much of it isn't clearly defined. The more you try to take from it the more the exact definition of what they're presenting matters.