This is an emergency use authorization, not a full approval by the FDA, and this only lasts until the state of emergency ends or the test is actually approved/rejected by the FDA. Either way, this is some exciting news as it could make sample collection easier for testing. It doesn't look like it overcomes the qPCR reagent bottlenecks to actually analyze the samples though, so I wonder what the full impact on testing will actually be.
And that's come with a lot of problems as we saw with the antibody testing fiasco with all the nonfunctional tests. So the precedent suggests cautious optimism, emphasis on the cautious.
Actually, emphasis on the optimism. The Yale work is really well done. Sensitivity is a little low, but people keep ignoring cost, scalability, frequency and ease. This is what is needed for back to work/school applications and the first test that actually satisfies these criteria imo.
Absolutely really good work. My concern is still that it doesn't seem to solve the qpcr bottleneck. Its great that it gets rid of ena extraction (as someone trying to get enough reagents to do any sort of real work in my lab - I'm all too aware of how much trouble they're saving themselves by cutting out that bit). The pcr part doesn't seem to be addressed though - amd if it isn't (without something like pooling or using something like LAMP) this will just add to the backlog of tests.
On the other hand, will still def second you on the quality of the work. It's some really solid science.
Main problem was that FDA set too high of a target for tests compared to PCR, iirc it was around 80%, where did they get that number is unknown. There were companies that had rapid antigen tests created back in February (fx E25Bio) which had lower sensitivity, so they were spending all that time to reach the goal, while also forced to make test more expensive. And main problem with that FDA mindset is that accuracy is not random, it depends on amount of viruses in the person and for people who were transmitting the decease accuracy approached 100%, and that's what we care about from the pandemic standpoint.
Tests can be properly regulated and checked by FDA, but if they set target with more thought behind it and acted quickly back in February and we had cheap 15 minutes tests, we wouldn't be in current situation now.
Isn't it still a problem if someone takes a test on day N and it comes back negative, but their viral load increases over the next few days and they begin spreading the virus on day N+3? Many people in such a situation would read the test as "I'm in the clear for a while" and might engage in close contact with friends and family members that would put them at risk.
To be clear, the problem with the antibody testing was that they were not monitored under the EUA until around April. Instead, manufacturers were permitted to market their tests without FDA review, provided that they told customers that the tests had not been reviewed.
They both work very similar in theory: unpurified saliva with qPCR. Though as far as I know U of I doesn't have their EUA and have not stated they applied.
On Aug. 14, 2020, Science Friday featured a discussion about rapid testing strategies with two scientists. The recording is well worth a listen. The key take-away is that while PCR tells you if you're infected, to protect others what you really want to know is if you're infectious.
A correct positive PCR doesn't even guarantee infection - it just tells you that you have been infected within the last month or so. Old, inactivated viral RNA can still trigger a positive weeks later.
Indeed, people testing positive months after all symptoms have disappeared don't appear to be infectious, the PCR is just picking up fragments of viral RNA.
That said, I always understood RNA as being pretty fragile, so I'm surprised, but I do know that the fragility is heavily dependent up on structure - you can have RNA sequences that are pretty stable due to the tertiary structure (folding) that occurs.
Are there examples of studies on people testing positive months later? qPCR is used in many settings for medical testing and the guidelines for other diseases are that residual bacterial RNA is cleared 7-14 days after successful antibiotic treatment. Obviously, COVID is caused by a virus, but a much longer period to clear viral RNA is not intuitive.
It still requires specific equipment- it looks like its still a qPCR based test, not an isothermal amplification type test, so it needs a qPCR machine and reagents. It cuts out one step, RNA extraction, which means it trades off some sensitivity for a faster and cheaper result.
Pretty curious to know what the false negative rate is. With the most popular current test (nasopharyngeal RT-PCR), the false negative rate is very roughly 30%, which is already pretty awful. And if you get tested significantly before or after peak viral load, it’s way worse than 30% [1]. If you get a negative result, you can’t really trust it, it’s giving false confidence to literally millions that they don’t have COVID, when they do.
It sounds like this test is not quite as good? If it has, say, a 40-50% false negative rate, I don’t really care how cheap and fast it is, that sucks. These tests are all pretty useless if you’re pre-symptomatic, so they’re not good for screening anyways, and if I’m at peak symptoms, I want accuracy over speed. I took a nasopharyngeal RT-PCR test and got results the same day, that’s fast enough if it’s more accurate. But if the accuracy is equivalent, or very nearly so, then it’s quite interesting.
> SalivaDirect is highly sensitive and yields similar outcomes as NP swabbing.
Doesn’t sound far better? And really, when hyping up a new technology, if I hear “similar outcomes”, my BS detector says “a bit worse.”
With NP swabs, the main issue doesn’t seem to be poor technique, it’s more that samples reasonably often don’t contain enough virus to be detected by the PCR test, unless you’re timing is really perfect (i.e. about 3 days after first spring symptoms). From my readings it’s not necessarily the collectors fault, there just didn’t happen to be high viral load in the part of the body when the sample was collected. Probably similar for saliva? Though I don’t really know.
This is interesting. I recently had a throat swab test after a week of extreme fatigue, but at no time did they indicate that the negative result was anything but certain. I had imagined _some_ false negative rate, but not this high.
This is why I laugh when people quote the infection numbers. It is potentially up to 30% wrong or heck, even more... nobody knows... but there is also other reasons, like inaccurate deduplication when people get multiple rounds of tests because of the high false negative rate.
Oh and because it is false negative, it is worse than you think, not better.
The preliminary study shows very high specificity, which is good for something that may be widely deployed, as it avoids the problem of massive numbers of false positives: https://lee-phillips.org/testingbad/
I'd like to see the FDA move away from approvals like this and just focus on lightweight approvals that "the test seems not fraudulent for presenting the false positive rate and false negative rates of the test, and it is indeed better than ransom".
Tests should not return "positive" or "negative", rather something like "3 in 4 chance of infection" of positive and "1 in 200 chance of infection" if negative (something like that). This is both more true and helps people understand statistics/how these tests work, instead of hiding it.
I don't see a safety risk for a test if it has some predictive value unless it is fraud.
There could be a separate threshold for "reimbursability" for insurance reasons.
We did this this morning, and it was not bad. They ship the tubes to your house, and you get in a Zoom call to verify ID and do the procedure. They haven't worked out all the kinks (it took much too long to do all of us) but it was pretty convenient. The biggest issue? It's surprisingly hard to fill up an entire test tube with spit.
From the article, it isn't clear how long it will take to get the results. Is it the same as the "traditional" test? Any ideas? Here in Oz, I received a negative in 24 hours (actually more like 14 ) using the nasopharyngeal method and which was free for me (covered by Medicare) so would love to know if the turnaround time for SalivaDirect will be lesser as well.
I'm unfamiliar with "pooled testing". I've read many universities plan on doing it with a saliva test. Is this something that's been done in the past? Frankly, it doesn't quite right to me but then again I'm not a scientist.
That‘s how China is able to test millions of people in a short amount of time.
It works only when you have low rate of infections.
Instead of testing each sample you test e.g. 5. If this pooled sample tests positive, you test each 5 individually again (for a total of 6 tests instead of 5). If the pool tests negative you‘re done and you‘ve only run 1 test instead of 5 individual ones.
So you can run the math yourself to see what % of infection rates it makes sense to pool and when not ;)
Without knowing the specifics (because it's not mentioned in the Yale press release), I can only guess. But it's something that we did in grad school to reduce sample processing costs looking for rare genetic variants.
The main idea is that if you are looking for a rare event, but testing an individual is expensive, time consuming, or difficult, then you can save some of that time and money by performing multiple tests at one time. So, instead of using a single test to see if a single person is infected, you'd use a single test to see if a pool/batch of people are infected.
Imagine a university with 20,000 students that has decided to test everyone once a week. Ordinarily, they'd need to run 20,000 tests a week. Now, assume they are going to use pooled testing. If your student population is 20,000 and the pool size is 100, instead of 20,000 tests a week, you'd only need to run 200 a week. 200 tests a week is manageable... 20,000 is not (from time, cost, and interpretation standpoints). The machines that run the qPCR test run in batches of 96 or 384 (not including controls), so you can see how reducing your total numbers of tests would be useful.
Pooling then answers the question of if anyone in a particular pool has Covid-19. If you have a positive signal from a pool, you should then go back and re-test all of the members of the pool. Even if you have multiple pools come back as positive and need to retest entire pools, you'd still be saving tests (and time/money). 20,000 vs 200 + 100 from each positive pool.
There are a few variations of pooling that can help make this even more efficient. If instead of 200 pools for 20,000 students, what if you had each student be part of 2 pools? Then you'd need twice as many tests per week (400), but if you have positive signals, then you'd only need to retest the students that were in both pools, which should be a very small number. If you went from having to retest 200 people to 5 people, then you're looking at even more savings. The requirement for 2 pools to have a hit is also a quality control measure -- what if you had a false positive in one pool, but no second hit from another pool... then you might assume that it was a false-positive and skip re-testing entirely.
This is only something that really works with rare events. But if you have a test with known false-positive and false-negative rates, then you can really save a lot with pooled testing... even with lower rates of pooling than 100. If you had pools of 5-10, you'd still be saving a lot of test reagents and costs. 20000 vs 2000-4000 is still something that would be a big advantage.
Pooled testing is currently done in Germany and has been done in many places.
It mixes samples and if the entire batch tests negative, you save yourself from having to run n tests. If it tests positive, then you do something akin to a “binary search” to efficiently narrow down the positive cases.
This is explained here. (See section on generalized binary splitting algorithm).
For me, it was quite painful. They stuck a swab to the back of my nose, which was uncomfortable. They they went several inches further to a place that I didn't even know I had. (I suspect it was the lower sinus cavity, but I'm not sure.) They jabbed it into the back of that, hard, which hurt. They left it there, pressed in hard, for 10 to 15 seconds.
On the upside, it felt like that sinus drained a bit after they pulled the probe out. It may have hurt me more because my sinuses are, um, fuller than they should be.
It's essentially the 'human blockhead' carnie trick. If you do it right it's strange and perhaps unpleasant, but not painful. The recommended method of learning how to do it is to use cotton swabs.
If you do it wrong, or maybe just have a particularly tight nasal passage, it could definitely be painful. For me, my right nostril is easier than the left.
Great news, Building on this idea. Is it also possible to calibrate an electric tongue to detect COVID-19 from a breath test? COVID-19 has a very specific smell I think.
Breath and saliva test would be quicker than PCR and would probably be important to protect the elderly as many who have COVID is asymptotic.
