Diagnostic tests balance sensitivity and specificity. This is done to balance false negatives and false positives. If you could design a testing protocol for Ebola, and you could use multiple tests, what would you do?
I would run a first test that is as fast as possible and is really good at saying as definitively as possible that a negative result means no Ebola. That would be a test with high sensitivity which means that if you test negative you are very, very unlikely to have Ebola.
I would then (or concurrently) run a test that doesn't include so many false positives. (This is referred to as high specificity.) And if I had another test that took longer but offered an even lower false positive rate I would add that into the mix.
It is a little more complicated than this, but I would take a real guess that the CDC is using a mix of tests like this right now. The real trick here is making sure that the combination of tests don't overlap in a way that causes problems if you are only blending high sensitivity or high specificity tests that are low on the other one. I doubt the CDC is worrying much about that right now as they rush to cover their posteriors.
Here is a link to better explain specificity and sensitivity. http://https://www.med.emory.edu/EMAC/curriculum/diagnosis/sensand.htm
To me, this presents a very reasonable way for them to say "no Ebola" right away for some and to take longer for others.
Slowplaying any positives is a whole 'another ball of wax, but from a technical standpoint this stands out clearly as a good reason for the different testing time frames.
----
To go just a touch deeper, if you had 3 tests going from a quick high sensitivity (say 99.9% sensitivity and 30% specificity), to a quicker high specificity test (50% sensitivity and 70% specificity), to a slower but more certain test to rule out false positives (80% sensitivity and 90% specitivity) you could get a whole chain of results.
Testing 1,000 suspected cases in the first round would give you 1 false negative and 300 false positives. You would release the negative results with a "no Ebola" statement but keep their info and run the other tests. With 700 false positives, you wouldn't be diagnosing anyone yet.
Test 2 would give you 500 false negatives and 300 false positives.
Test 3 would give you 200 false negatives and 100 false positives.
The special sauce is in interpreting this mess of results, and figuring out what overlap there is and what confounding variables. But even with all of that, the first quick test at least lets you clear the deck of a lot of people who don't have Ebola.
---
Please let me know any problems you see with this. My 2 2&under don't allow for much proof reading.
I would run a first test that is as fast as possible and is really good at saying as definitively as possible that a negative result means no Ebola. That would be a test with high sensitivity which means that if you test negative you are very, very unlikely to have Ebola.
I would then (or concurrently) run a test that doesn't include so many false positives. (This is referred to as high specificity.) And if I had another test that took longer but offered an even lower false positive rate I would add that into the mix.
It is a little more complicated than this, but I would take a real guess that the CDC is using a mix of tests like this right now. The real trick here is making sure that the combination of tests don't overlap in a way that causes problems if you are only blending high sensitivity or high specificity tests that are low on the other one. I doubt the CDC is worrying much about that right now as they rush to cover their posteriors.
Here is a link to better explain specificity and sensitivity. http://https://www.med.emory.edu/EMAC/curriculum/diagnosis/sensand.htm
To me, this presents a very reasonable way for them to say "no Ebola" right away for some and to take longer for others.
Slowplaying any positives is a whole 'another ball of wax, but from a technical standpoint this stands out clearly as a good reason for the different testing time frames.
----
To go just a touch deeper, if you had 3 tests going from a quick high sensitivity (say 99.9% sensitivity and 30% specificity), to a quicker high specificity test (50% sensitivity and 70% specificity), to a slower but more certain test to rule out false positives (80% sensitivity and 90% specitivity) you could get a whole chain of results.
Testing 1,000 suspected cases in the first round would give you 1 false negative and 300 false positives. You would release the negative results with a "no Ebola" statement but keep their info and run the other tests. With 700 false positives, you wouldn't be diagnosing anyone yet.
Test 2 would give you 500 false negatives and 300 false positives.
Test 3 would give you 200 false negatives and 100 false positives.
The special sauce is in interpreting this mess of results, and figuring out what overlap there is and what confounding variables. But even with all of that, the first quick test at least lets you clear the deck of a lot of people who don't have Ebola.
---
Please let me know any problems you see with this. My 2 2&under don't allow for much proof reading.