SoylentNews is people
SoylentNews
Testing half the population weekly with inexpensive, rapid-turnaround COVID-19 tests would drive the virus toward elimination within weeks—even if those tests are significantly less sensitive than gold-standard clinical tests, according to a new study published today by CU Boulder and Harvard University researchers.
Such a strategy could lead to "personalized stay-at-home orders" without shutting down restaurants, bars, retail stores and schools, the authors said.
"Our big picture finding is that, when it comes to public health, it's better to have a less sensitive test with results today than a more sensitive one with results tomorrow," said lead author Daniel Larremore, an assistant professor of computer science at CU Boulder. "Rather than telling everyone to stay home so you can be sure that one person who is sick doesn't spread it, we could give only the contagious people stay-at-home orders so everyone else can go about their lives."
[...] They then used mathematical modeling to forecast the impact of screening with different kinds of tests on three hypothetical scenarios: in 10,000 individuals; in a university-type setting of 20,000 people; and in a city of 8.4 million.
[...] When it came to curbing spread, they found that frequency and turnaround time are much more important than test sensitivity.
For instance, in one scenario in a large city, widespread twice-weekly testing with a rapid but less sensitive test reduced the degree of infectiousness, or R0 ("R naught"), of the virus by 80%. But twice-weekly testing with a more sensitive PCR (polymerase chain reaction) test, which takes up to 48 hours to return results, reduced infectiousness by only 58%. When the amount of testing was the same, the rapid test always reduced infectiousness better than the slower, more sensitive PCR test.
That's because about two-thirds of infected people have no symptoms and as they await their results, they continue to spread the virus.
"This paper is one of the first to show we should worry less about test sensitivity and, when it comes to public health, prioritize frequency and turnaround," said senior co-author Roy Parker, director of the BioFrontiers Institute and a Howard Hughes Medical Institute investigator.
Journal Reference:
Daniel B. Larremore, Bryan Wilder, Evan Lester, [et al]. Test sensitivity is secondary to frequency and turnaround time for COVID-19 screening. Science Advances, Nov. 20, 2020; DOI: 10.1126/sciadv.abd5393
(Score: 3, Informative) by Runaway1956 on Sunday November 22 2020, @04:07AM (23 children)
Rapid results makes sense, even at the cost of accuracy, up to a point. I mean, you can't have 1/4 of the population getting false positives every two weeks, and another 1/10 getting false negatives. At some point, you have to admit that it's all noise, without any meaningful signal. Soon - like real soon - people just stop cooperating.
Better keep working on the sensitivity.
Abortion is the number one killed of children in the United States.
(Score: 4, Insightful) by c0lo on Sunday November 22 2020, @04:30AM (20 children)
Homework: if the probability of getting a false positive is 10%, what is the probability after 7 tests to get at least one false positive**?
If a positive means "Go isolate yourself", you will end with a perpetual lockdown even after you eradicated the virus
--
**
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 2) by Runaway1956 on Sunday November 22 2020, @04:47AM (7 children)
Thank you for making my point with maths. ;^)
Abortion is the number one killed of children in the United States.
(Score: 2) by c0lo on Sunday November 22 2020, @05:26AM (1 child)
???
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 2, Informative) by hemocyanin on Sunday November 22 2020, @07:29AM
semi-colon to the close parens power is always funnier.
(Score: 5, Insightful) by Anonymous Coward on Sunday November 22 2020, @05:38AM (4 children)
Umm, your point was not proven. With 7 tests you barely go over 50% chance of a false positive, so if you get a positive test then quarantine and take a new test as soon as you can. It would be up to each area to figure out how many negatives tests would indicate your positive test is false.
Or did you not consider that it takes 7 tests to have a 52% chance of getting ONE positive result?
Also, thanks for pointing out why sometimes authoritarian measures are necessary. In a pandemic requiring a 2 week quarantine is hardly the height of tyranny, but if you're worried about missing work then perhaps you should go talk to your R reps and figure out why they are giving money to large corporations instead of people and small biz?
(Score: 2) by c0lo on Sunday November 22 2020, @06:03AM (1 child)
Assuming test being taken every 2 days and assuming you need at least two negatives to counter a positive, in a covid extinct situation you'd still have 50% of the population going into isolation for at least 4 days.
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 0) by Anonymous Coward on Monday November 23 2020, @10:24PM
The article talks about 1 test every 2 weeks (or half the population every week).
(Score: 2) by legont on Sunday November 22 2020, @03:08PM (1 child)
We leave in a zero tolerance society, remember? Once the infrastructure is in place, we will test for flu, hepatitis, aids and so on. Our "precious children" are still there to protect and even one "predator with a bug inside" is enough to take action.
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 4, Insightful) by Anonymous Coward on Sunday November 22 2020, @07:47PM
These are extraordinary conditions, and it bears repeating.
2-4 weeks of lockdown and quarantining of infected, then limited re-opening with full mask and social distance compliance would allow a near total re-opening with continued safety measures.
Assholes like you who treat masks like the end-times are fucking it up for everyone else. Stupid fucking kids holding huge parties then going home for large thanksgiving events are going to spread COVID to just about every community. Please, send us some more of your oppressed drivel.
(Score: 5, Interesting) by c0lo on Sunday November 22 2020, @07:02AM (10 children)
The results of a precise test can be obtained in 24h or less if the pressure on the labs don't overflow their capacity. Worked a charm in Melbourne case.
Took 111 days in which 75% of Melbournians stayed at home (and would have taken 2-3 weeks less if not for an idiot that lied the contact tracers and started two new clusters), but the results are wonderful - today's 3 weeks+ with 0 community cases and 0 deaths, we were announced no masks required outdoors if not in a crowd, restaurants allowing 150 people indoor (up from 30). We keep it like this for 1 more week, we get over the epidemiological threshold of 28 days and we'll likely have as close a normal Christmas as possible (i.e. keep an ear to the news, just in case some other unfortunate mistake let the virus escape from the repats quarantine).
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: -1, Troll) by Anonymous Coward on Sunday November 22 2020, @02:22PM (5 children)
here's by totally forgoten math:
so 10% is super. we can map outcomes to digits 0 - 9.
only one digit (10%) is interesting.
so let's map this case to the digit "1". (win!) all others map to lose.
now let's combine all possible combinations:
0000000
0000001
...
9999999
so we get 10 million combinations.
let's see how many have a "1" and only one "1":
1000000
0100000
0010000
...
0000001
so there are 7.
so 7 devide by 10 million is the probability for 10% in 7 series?
(Score: -1, Redundant) by Anonymous Coward on Sunday November 22 2020, @02:55PM
oops.
tbc ... remove one placeholder, so to account for "only one one" so 6 placeholders instead of 7.
also the digit "1" is not usable, only "9" digits are available so
count with "9" symbol to 6 placeholders
9^6 = 531'441
multiply all of 'em 531'441 unique symbol string made up of 9 unique symbols in 6 placeholders by 7 (7 places where a "1" could be and get 3'720'087.
(100 x 3'720'087) devide by all possible symbolstring with 10 symbols and 7 placeholders (10'000'000) and get ... 37.20087?
(Score: 2) by c0lo on Sunday November 22 2020, @05:04PM (3 children)
Why "only one 1"? Should've been "at least one 1".
And why all the others digits are suddenly mapped on "0"? Even if the outcome is the same, if you mapped probabilities to 9, then you must use it to compute the number of cases.
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 0) by Anonymous Coward on Sunday November 22 2020, @09:48PM (2 children)
i thought the original "homework" was rolling a 10 side dice with only 1 green side (all others being red) for a total of 7 times; what are your chances to get one and only one green.
anyways ... i was assuming no one wants green (not even once), cause that means you gotta stay home for two weeks?
so what's the correct number then?
(Score: 2) by c0lo on Sunday November 22 2020, @10:31PM
To get "at least one green". Go check.
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 0) by Anonymous Coward on Monday November 23 2020, @03:34AM
1 - (1 - 0.10) ^ 7 = 0.521703
ie. 52%
(Score: 0) by Anonymous Coward on Sunday November 22 2020, @09:15PM (3 children)
Meanwhile in Latvia, all summer new infections were in the single digits with NONE of that craziness. Magic?
(Score: 3, Informative) by c0lo on Sunday November 22 2020, @10:35PM (2 children)
Well, while Latvia [www.latvia.travel] were "all summer", we were "all winter". Magic?
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 0) by Anonymous Coward on Monday November 23 2020, @01:39AM (1 child)
Dude, the weather in what is called "winter" in Australia is a bit warmer than what is summer in Latvia (should you not know that already, as a professed ex-European?) It is rather to be expected for the respiratory illnesses there to peak differently.
Like in this report from 2019: https://globalnews.ca/news/5435232/australia-flu-season-canada/ [globalnews.ca]
"Usually, flu season doesn’t start until June or July, Booy said. This year, numbers started to pick up in mid-March."
Now, looking at Australian COVID peaks of 2020, I observe end of March and end of July. Isn't the nature predictable?
(Score: 2) by c0lo on Monday November 23 2020, @03:17AM
Do tell. Really? I wonder how I didn't notice. Perhaps because in winter Aussies gather in closed spaces too.
Go ahead, mate, your ability to predict nature should make you rich in no time.
https://www.youtube.com/watch?v=aoFiw2jMy-0
(Score: 3, Interesting) by maxwell demon on Sunday November 22 2020, @09:39PM
Not enough data for a meaningful answer.
You would need to know the correlation between those tests. I doubt that it is a purely random error; rather it will probably be some unknown condition of the tested person that will make the test a false positive, with the false positive rate measuring how many people have that condition. If that condition is one that is unlikely to change in the relevant time spans, then the probability of getting a false positive after having gotten a true negative will be close to zero. On the other hand, if that condition is likely to change between those test, the probability of testing positive will be almost unaffected by the results of the previous tests.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 1, Informative) by Anonymous Coward on Sunday November 22 2020, @05:28AM
Sensitivity does not measure false positives. The specificity of these tests is assumed to be 100%, but most tests used today are in that range so that assumption isn't unfounded.
(Score: 2) by krishnoid on Sunday November 22 2020, @06:14AM
And while the sensitivity is improved or combinations of tests are developed, in the meantime, it's less of a *public* health "surge" issue and more manageable with current ER/hospital capacity.