Here in North Carolina, we have 12 COVID 19 cases. Eleven of those cases are "presumptive positives," which means that a local test showed positive, but the CDC has not confirmed the result. Being facetious, the epidemic may be over before we get confirmation, but it really does hinder understanding for us average folks.
Something that's bandied about somewhat are the "sensitivity" and "specificity" of the tests. i'm far from being an expert on biology and epidemiology, but working in the life insurance industry during other epidemics required that i learn something about the mathematics of medical testing. For what it's worth, here's what I know.
If you go here?ÿ
https://www.biomedomics.com/products/infectious-disease/covid-19-rt/ ?ÿ and scroll down to "How accurate is the COVID 19 Rapid Test?" you can see some informative numbers. A total of 525 blood samples, 397 from clinically confirmed infected people and 128 non-infected people were tested.
Sensitivity and specificity refer to the probability that a test result is correct. Sensitivity is the probability that a sample that tested positive was actually infected. Specificity is the probability that a sample that tested negative was actually not infected. Typically for a given test, improving one impairs the other.
The table below summarizes the data for this test. Note that the incidence, the percentage of samples known to be infected, is quite high. Back in the mid-1980s when AIDS tests were being developed, one of the problems with validating the tests was the low incidence of disease in the general population. Mis-categorizing one sample could greatly alter the presumed accuracy of the test overall. What emerged was a two-test regimen: a highly-sensitive test (lots of false positives) used first and then a highly specific test to eliminate almost all of the false positives, leaving high confidence that the sample was indeed infected.
COVID 19 testing may be similar; I don't know.
Again, I'm not an expert, so please correct any mistakes you see. Developing reliable tests is not always easy, but the math can be helpful in understanding their limits.
Sensitivity is the probability that a sample that tested positive was actually infected. Specificity is the probability that a sample that tested negative was actually not infected. Typically for a given test, improving one impairs the other.
Not quite right - you give the correct associations for the terms, but the probabilities are conditional on the true value, not conditional on the test result.?ÿ You can see that in the percentages correctly calculated in the table, but the numbers don't work out to be the same if you reverse the conditionality.
Sensitivity is the probability that someone infected will test positive.
Specificity is the probability that someone not infected will test negative.
Here in North Carolina, we have 12 COVID 19 cases.
That you know of, probably a lot more in reality.
Sensitivity is the probability that a sample that tested positive was actually infected.
In the table, you can see that out of those who tested positive, 96.7% were infected. That's a useful number, but not the way sensitivity is usually defined.
Out of those who tested negative, only 72% were actually not infected.
There's a summary here of all the terminology, and as far as I checked it agrees with NIH. https://en.wikipedia.org/wiki/Sensitivity_and_specificity
Yep, I should have said, "Given that a sample is infected, what is the probability that the test is positive?"
Thanks for the correction to the reasoning.
indeed. The ACC tournament in Greensboro was cancelled today after two rounds were played on Tuesday and Wednesday. There likely was some virus spread on those first two days.
Of course, there was a bigger problem for Tar Heel fans. The Heels finished with a losing record for the season, but they could have made the NCAA tourney by winning the ACC tourney, thus becoming ACC champs.
Instead, it's Florida State, the regular season 1st place finisher.
The NBA has canceled all games until further notice...
Was it this bad during the Sars and Mers and Ebola scares? I don't remember all the "unprecedented" requirements being imposed. And it seems to me that those were a lot worse than this.
I'm no expert, so I'm probably not thinking about this correctly...
Was it this bad during the Sars and Mers and Ebola scares? I don't remember all the "unprecedented" requirements being imposed. And it seems to me that those were a lot worse than this.
SARS and MERS had a higher mortality among identified cases, but spread much more slowly so isolation of suspected cases was more effective without shutting down everything.?ÿ
Ebola was very effectively prevented from getting loose in the US because there was less spread before it was identified as a severe threat. People coming in were checked carefully, but it was business as usual inside the country.
The spread of viruses really deadly viruses like Ebola are actually hampered by their greater mortality rate; they kill their host too effectively to spread. I believe I read that the spread of either SARS or MERS was slowed because it wasn't communicable until a patient was symptomatic.
The "big one" will be something like this: high infection rate, long period of being communicable before a patient is systematic (week+), mortality rate of only 2% - 5%, but with the capacity to survive 24+ hours airborne.
Yep, we're up to 17 as of this morning, most about 70 - 80 miles from us, but 2 within 35 miles. At least 5 of the infected people attended the Biogen conference in Boston. Biogen has a 1000+ employee facility near Raleigh.
At this point I'm actually looking forward to getting the virus, getting over it, and moving on.
@norman-oklahoma
Me, too, although I'm in the "old" group with higher expected mortality. Angela Merkel said that she expects 2/3 of the German population will ultimately be infected. Maybe we'll do better, but it looks like a lot of people are going to get sick.
What I get from the numbers provided is that if the initial test says POSITIVE you had better get serious about treatment.?ÿ If the initial test says NEGATIVE you probably should be tested a second time per some better standard, if possible.?ÿ Must agree that the initial test is only accurate 89 percent of the time.?ÿ The small number of POSITIVES that should have been NEGATIVES does provide some hope however.?ÿ No need to immediately cash in all your assets, pre-pay the undertaker and immediately locate your burial spot.
From Twitter:
Back in the early days of AIDS testing, they used a two-test regimen. The first was a high sensitivity test, ELISA, that identified most of the positives, but also had a high false positive rate. Then those that tested positive by ELISA went through a high specificity test, the Western Blot. Because the ELISA positives represented a subset with a concentrated number of truly infected people, the high specificity test was appropriate. Whereas the initial ELISA tested population had maybe 4 or 5% truly infected, the subset that went to Western Blot had maybe 90-95% that were infected.
in the test above, about 75% of the people in the sample were truly infected. One of the problems with testing asymptomatic people is that the test may not work as well when there is a low incidence of disease. It's frustrating, but that's the way this stuff works
I haven't seen much about the CDC confirmatory test. It may be like the ELISA - Western Blot series or it may be something else. I have signed up for the W.H.O. daily updates and I look at CDC every day. The newspapers are doomsday, but the original source documents are informative.
However, pets can carry the virus on their bodies from an infected person to infect others.
