Duke Margolis researchers, along with Dr. Caitlin Rivers of John Hopkins University, have released a new issue brief on how we can improve our national testing strategy for COVID.
There are a couple of main points that we make.
- Develop Smarter Testing
The current testing regime is heavy on PCR tests which require skilled medical personnel, a complex lab and time. These tests are fairly expensive and potentially too slow to influence clinical and public health decision making. However, they are very specific and precise. If a PCR tests indicates that someone has COVID, they most likely do have COVID. This is needed for diagnostic purposes. We need more screening tests. These tests need to be faster and cheaper to run. The trade-off is that we should embrace more false positives as anyone who is screened as potentially positive should isolate while they receive a far more precise (albeit slower) PCR diagnostic test to get confirmation of either a true positive or cateagorize the screening result as a false positive.
Governor DeWine (R-OH) is a good example of how this system should work on a national scale. He was screened with a high speed, low cost antigen test that returned a positive signal. This shifted our prior on the probability of him being an non-symptomatic infected individual from unlikely to possible. And that change in priors led to a high value PCR test being used.
ETA: plot twist. The first test was apparently an antigen test, not a PCR test, and follow-up PCR test was negative. https://t.co/KE3kKE4Kis https://t.co/TSzDkCvp2O
— Dr. Tara C. Smith (@aetiology) August 7, 2020
At the same time, as he was waiting for the PCR test results to come back, he changed his behavior. He isolated himself. Some of his close contacts (his wife) also isolated themselves. If he had been infected, his actions significantly reduced the window in which he could have infected anyone else. He got the screening test because he was scheduled to meet with the President. That is not a sustainable model of national screening, but massive scaling up of cheap and accurate enough testing can break infection chains before they get started.
- Increase testing capacity
Congress should appropriate $45 billion dollars to create a national testing system that is focused on high risk populations (nursing homes, congregate living situations, schools. These tests don’t need to be single PCRs. It could be antigen or pooled PCR or a variety of other other technologies and approaches. Routine screening is plausible for locally (temporally and geographically) high risk populations for a year and a half.
- Widen the Supply Chain
Use the Defense Production Act to resolve immediate supply chain chokepoints. The Department of Health and Human Services (HHS) should publicly publish end to end supply chain analysis of all critical technologies and supplies on a regular basis. Furthermore, HHS should engage in advanced purchase arrangements and subscription deals (similar to Hep-C antivirals) to produce investment certainty.
- Tracing and Isolating
Testing is the first part of the public health response. Testing allows us to identify who needs interventions to break infection chains. Tracing and Isolating are our dominant public health interventions. $24 billion dollars should be provided to fund the costs of sufficient contact tracers to smother outbreaks before they start, pay for hotel rooms for people who need to be isolated but can not be safely isolated at home, and provide some, limited, wage/income support for individuals who need to be isolated but don’t have the ability to take sick time from work.
- Reporting
HHS should publish standardized and appropriately stratified reporting on key testing, and isolating indicators at the minimum state level if not county level. This will allow for more granular epidemic responses by public health authorities.
- Communicating
Clear, concise and consistent communication by trusted messengers is key. We need to know current testing status, turn around times for results, and contact tracing capacity at local, state and national levels.
syphonblue
Hey I think you broke the site by forgetting to close a [b]
WereBear
I think the key to this is HAVING ONE.
JaneE
I doubt anyone in the federal government will be concerned about improving any portion of our Covid-19 response before January 21. I would not be surprised to see the current administration do the exact opposite of any recommendations they have made.
BruceFromOhio
The model presumes that everyone is willing to get tested (they aren’t) and if tested positive, will change behaviors (they won’t). If it doesn’t have teeth, we’re stuck with whack-a-mole on outbreaks after the fact. See also: Dr Fauci, death threats.
Cheryl Rofer
David, it’s the Defense Procurement Act, but thank you for bolding this. It’s insane that Trump won’t do it while he’s waltzing around with other unnecessary EOs.
Dorothy A. Winsor
The DeWine thing did make me wonder about the reliability of the tests, so thanks for the explanation.
David Anderson
@BruceFromOhio: Don’t have to break every infection chain this way; just have to break more of them by either removing people upstream of a potential infection chain/super spreader event or minimizing the time that a person without symptoms is walking around and interacting with susceptible people.
It won’t stop everything, but it slows spread which is quite valuable.
Another Scott
Thanks for this, and thanks for your efforts. From the acknowledgement, it sounds like you should have been a co-author, but that’s for you folks to decide. ;-)
I hope this gets significant visibility and that it helps turn the tide in our national response so far. :-(
Cheers,
Scott.
p.a.
I dunno… none of this mentioned in 2nd Amendment…?
Haroldo
I don’t thank you enough Mr. Anderson, so…..thank you!
I learn a lot from your posts.
Another Scott
@Cheryl Rofer: I think it’s the “Defense Production Act” that’s applicable here.
https://en.wikipedia.org/wiki/Defense_Production_Act_of_1950
15 CFR §§700 to 700.93 – https://www.law.cornell.edu/cfr/text/15/part-700
;-)
Cheers,
Scott.
Fleeting Expletive
Is there, at this point, a cheap and simple spit-test that can produce more-reliable-than-useless results in minutes? ISTM that is what we’ll need, almost daily, for all students and school staff in order to open schools without killing children.
Gov. Stitt of Oklahoma now says he caught the ‘Rona from hugging dear old friends.
What can I say, that’s how we grow ’em hereabouts.
Carlo
I’m a little uncomfortable with the idea of rolling out inaccurate antigen tests at scale. I understand the logic, but I think this plan underestimates the likely public backlash when tens of thousands of false positive results are issued.
Pooled RT-PCR seems like a better option to me. Of course you still need to invest in test kits and lab capacity on a massive scale to make it work, but we need that anyway.
bbleh
“The fish rots from the head.” There will be no coordinated strategy, no massive allocation of resources, not even an acknowledgement of the problem, while Trump is president. It won’t be allowed to happen.
Ken
@Dorothy A. Winsor: The DeWine thing also makes it clear that everyone who’s meeting Dear Leader is being tested, and not admitted to the Presence if the test is positive. Which seems like an overreaction to a hoax virus, that isn’t any worse than the flu, and only kills people with comorbidities which Dear Leader doesn’t have, and doesn’t even kill them since they’re just dying with it not of it.
I’m also wondering if all the big-pocket donors who went to that yacht-club fundraiser were tested.
frosty
@bbleh: Yes. Every time I read “should” in the post I shook my head, “Nope”.
Mary G
Hillary would’ve done all this in February, so there’s no way in hell any Republicans will do it.
MomSense
Here in Maine the Mills Administration teamed up with IDEXX Labs, makers of veterinary tests, and they are providing testing for us. Apparently our State CDC Director and HHS Director reached out to them very early (January?) so we do have better testing here than in other places.
MomSense
@Carlo:
My understanding is that the pooled RT-PCR tests have high false negatives. Is that the case?
RepubAnon
@Ken: Yes, isn’t it odd that Trump wants to slow the testing down for everyone else… but insists on thorough testing for anyone visiting him.
Omnes Omnibus
@RepubAnon: No, it isn’t odd at all.
Cheryl Rofer
@Fleeting Expletive: I saw a journal article yesterday that one of the experts I follow seemed to think was a good basis for a spit test. When we can do those at home, we’ll be in much better shape.
Mokum
@Fleeting Expletive:
The university of Illinois at Urbana-Champaign says they have new saliva test which should be very quickand about as reliable as the standard test. They plan to test 10,000 people a day, everybody will be tested twice a week, 50,000 students and 20,000 faculty/staff. They repurposed a veterinary lab for this. It sounds all widely optimistic, but who knows.
this a preprint proposing the system https://www.biorxiv.org/content/10.1101/2020.06.18.159434v1
raven
@Mokum: OSKEEWOWOW!
Raoul
@JaneE:
And the GOP Senate is walking the plank with Trump.
I keep seeing these blind items about how ‘vulnerable’ GOP members up in 2020 are freaking out in private. And I just laugh (with bitterness).
Private carping will do nothing. If you cannot address your constituents suffering, then it’s obvious you also won’t address the shambles that is our federal ‘response’. It is a crime against humanity to have to wait till January to begin to build the systems David lists in his OP. But, dammit, here we are.
Vote ’em out, and look at prosecuting the culpable.
l3000
I just read an article somewhere about testing strips that could be done daily. They would allow for the same kind of immediate, temporary, quarantine while waiting on more precise test to validate the at home result. I think it was being developed by someone at Harvard? The idea would be to send large quantities of test strip packets to every household to use at home and/or workplace to provide fast, daily system of testing. Comparison with pregnancy tests available at grocery–you’d do a fast, screen test at home, then go to doctor for precise test to validate results–in the interim you’d have ‘evidence’ to stay home for work and protect yourself and your family quickly. Any idea if these are really accurate or feasible? Thanks for this post and ref.
Guess I should have read all the comments first. This must be same as spit tests mentioned already.
Raoul
@Mokum:
Just had a video happy hour with friends last evening, their eldest son has flown east to start his second year of college in MA (I honestly don’t recall which college, not an Ivy but a good mid-tier).
He’s (erm) in lockdown with his girlfriend for 14 days — so they decided she’d take that modest risk as he needed a landing place given MA rules, the other choice being 14 unaffordable hotel nights — before moving to a dorm.
The school will be doing at least weekly testing, a roughly 50/50 hybrid in-person (more in person, split class cohort for lab type courses, more online for lecture), and will have an entire dorm just for quarantine.
The parents are a little freaked out. If #1 son does test positive, the only option is to whisk him into the lockdown dorm for at least 14 days, and trust that the school’s health system will monitor his status and figure out elevated care if he should need it.
I just can’t imagine. But it did sound like this school was working pretty hard at it. Not like public high schools in, say, Georgia with their ‘what, me enforce masks?’ utter, contemptible bullshit.
eta: weekly would be much easier with a spit test!
Raoul
Deleted — strange — BJ manufactured a dupe post
VeniceRiley
As with Sturgis, I don’t trust my fellow humans to be responsible. I more expect behavior along the lines of “OMG I have it. Better catch a flight home before I start showing symptoms and have to isolate.”
So any roll out of fast saliva or prick tests to schools and airports would be a godsend, at least until we have vaccines or a treatment for mild to moderate cases.
I’d very much prefer faster trials of some of the treatments that are promising. Seems like everything is being trialled in hospitals and nursing homes. Can we just have 30,000 real inhalers of varying treatment formulations and 15,000 fake ones handed out and see what happens in a couple weeks? Jesus.
Haroldo
On Wisconsin:
https://news.wisc.edu/simpler-covid-19-test-could-provide-results-in-hours-from-saliva/
Carlo
@MomSense: I believe that pooled RT-PCR as an assay has the same error rate – sensitivity and specificity – as individual sampling. If there is higher error (I have not seen that discussed) it would have to be due to sample preparation problems, and I believe that would be addressable by protocol tuning.
Another limitation is that pooling is only cost-effective when the test-positive rate is low, since you need a fair fraction of the pooled samples to come up negative. In Florida now that might become an issue, though again addressable by tuning pool batch sizes.
Ultimately pooling is a good option for the kind of unstable equilibrium situations we see in Europe and Asia – low test-positive rates, frequent local outbreaks.If we could also reach that stage in the U.S., I believe it would work better than antigen screening to streamline the pipeline.
debbie
And should he turn out not to have COVID, DeWine has managed to increase the numbers of test skeptics, masking skeptics (a major problem around here), and social distancing skeptics. He’s managed to give his opposition additional weaponry to fight him and his policies. This isn’t good news for Ohio.
Villago Delenda Est
Biden might listen to this, but Donald will not. So until January 20th (we hope) we’re fucked.
BruceFromOhio
@David Anderson: I am totally onboard, and some is better than none. And we can most certainly do a better job of providing needed resources to the first-line providers.
I’m stymied by the apparent recasting of America as a death cult.
Meyerman
The good folks at This Week in Virology (https://www.microbe.tv/twiv/) have been plugging the use of rapid antigen tests for a while. They scoff at the idea of reopening schools without regular testing in place.
piratedan
now all you have to do Mr. Anderson is convince insurance companies to pay twice for what many in that industry will perceive to be the “same test”.
good luck with that….
a huge part of this is on big pharma, which will undoubtedly try to market them as the “same test” and demand to be paid an equivalent markup
BruceFromOhio
@debbie: The skeptics I’ve encountered we’re gonna skeptic anyway, and other than saying “Fuck it, everything is open with no restrictions!” the governor can’t do anything acceptable. Good on DeWine for doing the right thing, or else risk ending up like Herman Cain. I’d rather he be right and alive, than sick or dead and popular with covidiots. Fuck those assholes and the conspiracy theories they fill their FB and Twatter streams with.
Yutsano
@Haroldo: Sent to a Badger. He’s seen a couple of them. Not sure what good it will do for him as he tested positive for the antibodies. But maybe he could volunteer just to see what results he gets.
JR
At the very least there should be a widespread rollout of colorimetric isothermal assays (RT-LAMP). There are credible manufacturers like NEB involved and the cost is negligible. Takes 1/2 hour and only requires an incubator
cain
I can’t wait for Joe Biden to be elected president so that we can do these things.
cain
So you do what we did in WW2. You essentially declare war against this virus – and then start pumping the propaganda out. Essentially, accusing anyone not wearing a mask, not following the guideline as un-patriotic.
These folks will scream and holler – don’t care. Arrest their asses if they don’t follow it. This is a national emergency – not only from a health, but from a defense perspective. We can’t have our citizens working against the nation’s interests.
It’s important to project and show strength. That’s the only way to show these idiots that we all mean business.
Mokum
@raven: yeah, the Chief forever, right?
Fleeting Expletive
@Mokum: Thank you. I thought I saw something like that. Few news outlets seem to say much about Oxford’s vaccine development nor mitigation strategies being used in Europe and elsewhere.
Mokum
@Raoul: some math modelers at the university of Illinois found that once a week testing would not be anough, at the end of the semester there would be 35,000 positive cases. Twice weekly would keep it “under control”, I think 500 cases. Butthat seems to indicate that the university is okay with one or two deaths per semester. I am happy that I am recently retired and am not responsible for suchdecisions.
Roger Moore
@WereBear:
This x1000. The closest we currently have to a national testing strategy is “testing is bad because it makes Trump look bad”. Any halfway serious strategy would be a massive improvement.
Roger Moore
@David Anderson:
Yep. All we need is to get R well below 1, and we’ll get the virus squashed reasonably quickly. The key is that the better squashed the virus is the easier it is to implement less restrictive policies like test and trace rather than broad shut-downs, and the easier it is to get the virus back under control if there’s a flare-up.
trollhattan
Heard an interview this a.m. with an epidemiologist who recommends for colleges reopening to institute testing of EVERYBODY every two days. Which I presume zero US colleges are planning or can afford. My kid’s freshman year start date was just pushed back a week and I strongly suspect they’ll cancel on-campus housing completely (they’ve already reduced capacity by nearly half by making all rooms and suites single-occupancy). She’s beside herself with all the uncertainty, as she’s also supposed to be running, one of many fall sports on a bubble.
Ugh.
trollhattan
California policy-makers must feel rather stymied right now, given this ongoing problem:
Notice appears atop our county’s COVID page and makes the apparent dramatic drop in new cases suspect. Since it doesn’t affect fatality stats, that is at or near its all-time high, but deaths lag behind new cases so we really don’t know if current measures are effective or not.
Roger Moore
@Ken:
It’s clear that Trump and/or the people looking after him understand the importance of testing, tracing, and isolating in keeping the virus under control. They just don’t care enough about the rest of us to bother coming up with a strategy to give ordinary people the same kinds of protection they get. It’s the perfect encapsulation of Republican policies.
Haroldo
@Yutsano:
Let’s hope it’s of some use.
raven
@Mokum: Nah, I really had no problem with his retirement.
joel hanes
Hypothesis consistent with both of DeWine’s test results:
He had an asymptomatic case some time ago, and recovered.
That would produce a positive antigen detection, and a negative PCR test — which is what he got.
Brutusettu
That 1st positive test for DeWine set off a wave of conspiracies theories.
Martin
I’ll add a few things to this.
First is to work toward standard tests that can be handled by multiple labs. That way demand can move to where supply is available.
The existing testing infrastructure should be prioritized to front line workers. Basically anyone the state deems as essential. If there’s not enough capacity to treat essential workers, then you’ve given too many people permission to call themselves essential. Those two elements need to be tied. You want schools to reopen? Show us the testing capacity for all of those teachers, administrators and students.
Policies on test processing need to be improved. Tests need to be reported back to the individual within 24 hours. Any tests older than that get thrown away. I know that sounds harsh, but if you have a testing system which is falling behind, it will stay behind, and timeliness is critical. Your capacity is what you can do in 24 hours, not how many tests come in the door.
Switch the public tests to spit tests and take 3 samples per patient. You can then radically speed up testing by combining samples and finding positives through cross-referencing. Let’s take New York as a good example. Right now they’ve got about a 1% positivity rate. So they’re running 100 tests to find one positive. That’s pretty inefficient. But what they could do is with a 1% positivity rate, combine one of each sample with 14 others and just run one test. Then, combine a second of each sample with a different 14 others. There’s a 95% chance that there will only ever be one positive result in each batch, and if two tests come back positive, you look at which was the unique sample present in each. That’s almost certainly your positive case. And you can then use the 3rd sample to verify it. In the rare chance you get two positives in the same batch, you can again narrow it down, but you can then just test all 15 3rd samples. You’ll only need to do that 5% of the time.
So, with 1000 patients, you would run 67 tests (the first combination of 15 samples) plus another 67 tests (the second combination of 15 samples) plus 10 (the 3rd samples for the 10 positives you find) plus 60 (the 4 situations where a batch had two positives and you need to test all 15). So instead of 1000 tests, you’re only running at most 204. That’s a 5x improvement.
Now, since you just increased throughput by 5x, you can administer 5000 tests instead. Ideally that’ll drop your positivity rate to 0.2%, assuming all of the new tests are negative (this is the school situation where you are now testing 100% of the population instead of just those who think they might be sick). At 0.2%, you can up your sample size to 30, still have the same accuracy, and still come in under your capacity allowing you to increase again. You create a positive feedback loop where the better the positivity rate, the more tests you can process with the same manpower. In a few hours you can write a program to take the previous days positivity rate and calculate how many tests you can handle, how many samples per batch, tell you how to construct each batch so every sample is used twice and never with the same other samples, and how to find the positive cases among the results.
Now, the spit test may not be super accurate, but that’s okay. What you’re doing is quickly triaging a larger and larger portion of the population. Spit tests are super fast to collect compared to nasal swabs. And with a quick turnaround on results, you can then go back to those presumptive positives and follow up with a more accurate test, which you’ll certainly have capacity for because it’ll only be 1% the size of what you were doing before.
But, this requires coordination. It also requires directing the labs to change what they’re doing. It also requires someone to deal with the ‘hey, we now need millions of spit test sample containers, labeling and serializing for batches of 3, etc.’ That’s got to be the feds, and the feds are still in the ‘lets blame democratic governors for everyone dying’ phase of this.
But if I was in Bidens transition team, I’d be lining this up right now, talking to vendors, getting them ready for an avalanche of executive orders and defense production act demands on 1/20 and talking to states to set up the local framework of how to ensure every essential worker gets tested, and how to scale up for public testing.
Martin
@trollhattan: Actually, quite a few colleges can do that – at least research universities. We can process thousands of tests per day but are currently processing zero because we can’t feed data into an EPIC system. But we can do our own population.
Our plan was weekly tests, not every 2 days. Modeling indicated that with other measures in place, weekly should suffice, with Wed-Thurs being the optimal day given that new infections were most likely to happen over the weekend and there is an incubation period before which testing doesn’t catch cases. With some additional staffing, and assuming we turn off everything else those labs were doing (not so easy) we could handle the entire university given the kind of testing protocol outlined above. Assuming almost nobody is positive, batching tests should work well and require relatively little capacity.
Martin
@Mokum: There’s a real debate as to what the standard should be. We’re taking a more proactive view that so long as we keep the infection/fatality rate below what it would be if students didn’t return to campus, that’s a success.
But if your goal is zero, that requires a lot more testing.
Our view is that university is not supposed to remove all risk from life. We have students die every year, and it’s tragic, but we’d have as many or more die each year if they weren’t at university. We can’t protect everyone from everything, but we do need to demonstrate we are a safer place to be than to not be.
Weekly testing *when combined with other structural changes to reduce transmission* should suffice. I’m seeing very few effective structural changes being offered at other universities though. Putting up a sneeze guard on the instructors podium doesn’t cut it.
Bill Arnold
@Martin:
Let’s take New York as a good example. Right now they’ve got about a 1% positivity rate. So they’re running 100 tests to find one positive. That’s pretty inefficient. But what they could do is with a 1% positivity rate, combine one of each sample with 14 others and just run one test. Then, combine a second of each sample with a different 14 others.
As a New Yorker, I would very much like to see this, if it improved time to test results.
I’m traveling to Vermont in a few weeks, and because my county in NY is above the Vermont threshold, they demand either a test plus 7 days quarantine prior to visiting or a 14 day quarantine. (This is for lodging.)
Unfortunately, all the local test facilities are quoting COVID-19 PCR test results in 7-10 days, and 10 days is nearly two weeks.
Also many people will have cleared the virus within 10 days after a test (e..g. if they got it after a few days of symptoms) so it’s close to useless.
EthylEster
@BruceFromOhio: you quoted frontpager “At the same time, as he was waiting for the PCR test results to come back, he changed his behavior. He isolated himself. Some of his close contacts (his wife) also isolated themselves.”
The time between the first and second test MUST be short if this is going to work. DeWine has followup test results in a day, no? But many here and elsewhere have complained about having to wait a week for a test result.
Does anyone know WHAT is causing the lag in PCR test results now? Is it a backlog that only little people experience? Can Quest not hire more people to run the samples? Are ancillary reagents in short supply?
Tony Gerace
@p.a.: I, for one, plan to shoot that Chinese virus with my AR-15. Freedumb!
artem1s
my immediate question about all of this was the same testing protocol used for the Tulsa plague rally? Does the antigen test also return false negatives at a high rate? Did Herman Cain get tested before the Tulsa rally and have a false negative? Was his behavior affected negatively because of a false negative (we can assume the answer is ‘yes’. he would insist on being an unmasked ass no matter the tests results). What if there were also false negatives for last night’s event? I think we can safely assume this is a similar crowd to those who would have had up close access to Trump in Tulsa. How many GOPers came to this event with the plague and went home last thinking they were all safe? And worse, will now mock DeWine for being such a pansy, mask wearing science believer?
Sab
@David Anderson: Yes for breaking even some infection chains.
The way DeWine did it ( and thanks to him for being open about what is going on) he didn’t have to do a full 14 day quarantine with a false positive. Instead a better test second time around reduces his quarantine to something reasonable.
In NE Ohio they are currently refusing to do any testing of kids under 18 until they are hospitalized. Yet some schools ( mostly parochial) are opening and having sports. So now what if a kid gets mildish Covid symptoms? 14 day quarantine for whole family? Assume it’s something else? Ignore it?
You can’t have whole families with school kids on 14 day lockdown every time anyone has a fever. Parents would be unemployed within months. But the alternative currently seems to be do nothing, and anyone who can work from home and does online schooling just opts out of any communal life, and tough luck for everyone else. What a way to run a society.
Raoul
@Mokum: I think maybe the parents did say twice a week? We talked about a lot of other things last night, it could also have been once every two weeks? I’m really not sure, it all sounded kind of horrible to me. My WFH bubble feels vary calm by comparison.
Sab
@Sab: Also too with sports. Heard a Cleveland public school guy being interviewed. He wants to open school sports, but he said most of kids are higher risk minorities. Most of their parents work multiple essential worker jobs. Many of parents and kids have other health issues like asthma, diabetes, lead poisoning, etc.
They wanted to postpone football until Spring and just play intracity, but OHSAA would penalize them for training out of season. Urk bureaucrats.