COVID: "low" transmission etc, is that daily or weekly cases?
April 5, 2022 6:05 PM Subscribe
from an article about CDC transmission categories:
"A county's level of transmission is based on just two metrics: new Covid-19 cases per 100,000 people and the positivity rate, both measured over the last seven days. The basic idea is that these show how much virus is spreading around us, CNN Medical Analyst Dr. Leana Wen said.
"The level of coronavirus transmission is an important determinant of how much risk you're in, if you're vaccinated or unvaccinated," she said.
Specifically, a "Low" transmission is considered no more than 10 cases per 100,000 people, or a test positivity rate of less than 5%. "Moderate" transmission is 10 to 50 cases per 100,000 people, or a positivity rate between 5% and 8%. "Substantial" transmission is 50 to 100 cases per 100,000, or a positivity rate between 8% and 10%, and "high" transmission is 100 or more cases per 100,000 people or a positivity rate of 10% or higher."
What do they mean by "measured over the last seven days"? Is that the seven-day average? Or is that TOTAL number per 100,000 over 7 days?
"The level of coronavirus transmission is an important determinant of how much risk you're in, if you're vaccinated or unvaccinated," she said.
Specifically, a "Low" transmission is considered no more than 10 cases per 100,000 people, or a test positivity rate of less than 5%. "Moderate" transmission is 10 to 50 cases per 100,000 people, or a positivity rate between 5% and 8%. "Substantial" transmission is 50 to 100 cases per 100,000, or a positivity rate between 8% and 10%, and "high" transmission is 100 or more cases per 100,000 people or a positivity rate of 10% or higher."
What do they mean by "measured over the last seven days"? Is that the seven-day average? Or is that TOTAL number per 100,000 over 7 days?
I agree, CDC Case rate = Cases / Last 7 days / 100,000 population (not Cases/Day/100k)
But that's the old rules.
The new (2022 March 24) rules are much *much* more lenient:
https://www.cdc.gov/coronavirus/2019-ncov/science/community-levels.html
Under the new rules, anything below 200 cases per 100,000 per week is "green" and "low" unless the sky is falling.
posted by soylent00FF00 at 6:56 PM on April 5, 2022 [1 favorite]
But that's the old rules.
The new (2022 March 24) rules are much *much* more lenient:
https://www.cdc.gov/coronavirus/2019-ncov/science/community-levels.html
Under the new rules, anything below 200 cases per 100,000 per week is "green" and "low" unless the sky is falling.
posted by soylent00FF00 at 6:56 PM on April 5, 2022 [1 favorite]
Response by poster: thank you. I understand that the rules have changed and I appreciate your pointing that out, but what I was confused about was the "daily" vs "weekly". I'd always thought that the old "no more than 10 per 100,000" was daily. Now I guess they meant weekly, so fewer than 2 new cases per day per 100,000 -- which seems like very very few people (very stringent standards). Yet now it's "loose" standards.
posted by DMelanogaster at 7:01 PM on April 5, 2022
posted by DMelanogaster at 7:01 PM on April 5, 2022
200 cases per 100k per week = about 30 cases per 100k per day, by my math. (200/7 = 28.57 or so).
posted by Alterscape at 7:04 PM on April 5, 2022
posted by Alterscape at 7:04 PM on April 5, 2022
Not to get too political on this, but Dr. Leana Wen seems to have joined the "urgency of normal" camp, at the minimum amplifying their message about ending all restrictions. Dr. Wen at least gave a bit more "reasonable" approach, with a good dose of bothsided-ness, but she was quoting Dr. Lucy McBride, who was calling restrictions like masking and vaccine mandates "COVID theater". Of course, Dr. Wen didn't use the more "extreme" views of McBride in her opinion piece on WaPo.
Please also keep in mind that now COVID tests may no longer be free, testing rates are going to drop, and that's going to affect positivity rate. According to Becker's Hospital Review as of today, Alaska has stopped testing for COVID, and somehow Alabama is getting 100% positivity despite 122.8 tests per 100K, and Colorado has an alarming 63.8% positivity rate (all 7-day moving averages, the 0's are probably data glitches)
posted by kschang at 7:38 PM on April 5, 2022 [11 favorites]
Please also keep in mind that now COVID tests may no longer be free, testing rates are going to drop, and that's going to affect positivity rate. According to Becker's Hospital Review as of today, Alaska has stopped testing for COVID, and somehow Alabama is getting 100% positivity despite 122.8 tests per 100K, and Colorado has an alarming 63.8% positivity rate (all 7-day moving averages, the 0's are probably data glitches)
posted by kschang at 7:38 PM on April 5, 2022 [11 favorites]
Yep ... before, low transmission criteria was: no more than an average of 10 new cases per per 100,000 per week (which would average less than 2 per day per 100,000). So there could have been 0 cases Sunday through Thursday, and, say, 9 cases on Friday, but the criteria of no more than 10 per week would still have been met.
The new criteria is: no more than 200 cases per 100,000 per week (which would average about 28 or 29 per day per 100,000). And added to the new criteria is: less than 10 covid hospital admissions per 100,000 people over the past week (which would average less than 2 hospital admissions per day per 100,000). And, like the above, there could have been 0 hospital admissions Sunday through Thursday, and, say, 9 hospital admissions on Friday, but the criteria of no more than 10 per week would still have been met.
posted by SageTrail at 8:01 PM on April 5, 2022 [2 favorites]
The new criteria is: no more than 200 cases per 100,000 per week (which would average about 28 or 29 per day per 100,000). And added to the new criteria is: less than 10 covid hospital admissions per 100,000 people over the past week (which would average less than 2 hospital admissions per day per 100,000). And, like the above, there could have been 0 hospital admissions Sunday through Thursday, and, say, 9 hospital admissions on Friday, but the criteria of no more than 10 per week would still have been met.
posted by SageTrail at 8:01 PM on April 5, 2022 [2 favorites]
Yes, "weekly" probably means a 7-day running average. I never seen cases per week, but keep an eye on context because someone might do it sometime.
It could be described in a bunch of different ways, but I think it could be said that the CDC is grading on a curve, adjusting the scale as seems likely to be most helpful. There is no point in publishing a map with one state green and the other 49 red. My impression has been that more than 10 cases per 100,00 is unstable, and will lead to a local surge but apparently that's just what's happened where I live and/or the dynamic has changed in the balance of transmission vs vaccination.
posted by SemiSalt at 5:52 AM on April 6, 2022
It could be described in a bunch of different ways, but I think it could be said that the CDC is grading on a curve, adjusting the scale as seems likely to be most helpful. There is no point in publishing a map with one state green and the other 49 red. My impression has been that more than 10 cases per 100,00 is unstable, and will lead to a local surge but apparently that's just what's happened where I live and/or the dynamic has changed in the balance of transmission vs vaccination.
posted by SemiSalt at 5:52 AM on April 6, 2022
As to why 7 days instead of 1 day - I think it makes a certain amount of sense, in that people are contagious with COVID for roughly one week. Using the 7-day number then lets you answer the question "(roughly) what percent of people have COVID right now".
That's my theory at least, I've never seen this written. The 7 day rate also smooths out the daily fluctuations.
posted by soylent00FF00 at 7:14 AM on April 6, 2022
That's my theory at least, I've never seen this written. The 7 day rate also smooths out the daily fluctuations.
posted by soylent00FF00 at 7:14 AM on April 6, 2022
Response by poster: Even after reading all these answers, I'm not sure if this 7-day average means, for example:
Mon - 7 new cases
Tues - 5 new cases
Wed - 5 new cases
Thur - 4 new cases
Fri - 6 new cases
Sat - 4 new cases
Sun - 4 new cases
Is the "over the seven days" referred to in the article I quoted from (from CNN, by the way) 7+5+5+4+6+4+4 = 35/7 = 5 per 100,000? (because we are talking about how many new cases appeared each DAY, on average? as calculated by looking at each day's new cases for a week?) and we're supposed to assume that they're talking about new cases PER DAY, even though they're using the week's DAILY cases and averaging them to get a better number?
OR
Is the "over the seven days" 35 per 100,000, because, well, that's how many new cases appeared that WEEK, and "over the seven days" means PER WEEK?
Can we say for sure?
I look frequently at the case numbers on gothamist.com. They get their data from the New York City Board of Health. Their top graph shows a DAILY count, but if you scroll down the page there is vaccinated vs unvaccinated and "citywide" (both, weighted), and that is clearly a WEEKLY number (I think).
It's amazing that this is so unclear. No, actually, it's not. All of this is unclear.
posted by DMelanogaster at 10:02 AM on April 6, 2022
Mon - 7 new cases
Tues - 5 new cases
Wed - 5 new cases
Thur - 4 new cases
Fri - 6 new cases
Sat - 4 new cases
Sun - 4 new cases
Is the "over the seven days" referred to in the article I quoted from (from CNN, by the way) 7+5+5+4+6+4+4 = 35/7 = 5 per 100,000? (because we are talking about how many new cases appeared each DAY, on average? as calculated by looking at each day's new cases for a week?) and we're supposed to assume that they're talking about new cases PER DAY, even though they're using the week's DAILY cases and averaging them to get a better number?
OR
Is the "over the seven days" 35 per 100,000, because, well, that's how many new cases appeared that WEEK, and "over the seven days" means PER WEEK?
Can we say for sure?
I look frequently at the case numbers on gothamist.com. They get their data from the New York City Board of Health. Their top graph shows a DAILY count, but if you scroll down the page there is vaccinated vs unvaccinated and "citywide" (both, weighted), and that is clearly a WEEKLY number (I think).
It's amazing that this is so unclear. No, actually, it's not. All of this is unclear.
posted by DMelanogaster at 10:02 AM on April 6, 2022
Is the "over the seven days" 35 per 100,000, because, well, that's how many new cases appeared that WEEK, and "over the seven days" means PER WEEK?"
Speaking solely for myself, I can't see any other way to interpret "Number of positive tests in the county... during the past 7 days divided by the population in the county " than this. If you take it literally, they are taking the number of positive tests collected over the span of a week and dividing that number by the county population.
If the CDC are doing anything more involved (such as averaging the daily numbers, etc), the caption on that table would be wrong.
posted by BungaDunga at 10:55 AM on April 6, 2022
Speaking solely for myself, I can't see any other way to interpret "Number of positive tests in the county... during the past 7 days divided by the population in the county " than this. If you take it literally, they are taking the number of positive tests collected over the span of a week and dividing that number by the county population.
If the CDC are doing anything more involved (such as averaging the daily numbers, etc), the caption on that table would be wrong.
posted by BungaDunga at 10:55 AM on April 6, 2022
Relatedly: with respect to positivity rates, averaging daily rates is almost certainly the wrong thing to do.
If you have a weird day with a low number of tests, it will still contribute 1/7th of the final rate. Imagine a totally pathological case where exactly one test is done on Monday and it's positive, and a thousand tests were done the other six days and 100 were positive on each of those days. The positivity rate should be something like 1% but instead it will be (100% + 1%*6)/7 = 15%. That's obviously wrong! So they're definitely not doing that with the 7-day positivity rate.
posted by BungaDunga at 11:03 AM on April 6, 2022
If you have a weird day with a low number of tests, it will still contribute 1/7th of the final rate. Imagine a totally pathological case where exactly one test is done on Monday and it's positive, and a thousand tests were done the other six days and 100 were positive on each of those days. The positivity rate should be something like 1% but instead it will be (100% + 1%*6)/7 = 15%. That's obviously wrong! So they're definitely not doing that with the 7-day positivity rate.
posted by BungaDunga at 11:03 AM on April 6, 2022
This thread is closed to new comments.
The former is calculated as "Number of new cases in the county (or other administrative level) in the past 7 days divided by the population in the county (or other administrative level) multiplied by 100,000."
The latter is calculated as "Number of positive tests in the county (or other administrative level) during the past 7 days divided by the total number of tests performed in the county (or other administrative level) during the past 7 days."
posted by BungaDunga at 6:33 PM on April 5, 2022