This story first appeared at ProPublica. ProPublica is a Pulitzer Prize-winning investigative newsroom. Sign up for The Big Story newsletter to receive stories like this one in your inbox.
Excellent article. The only other thing to add would be the specificity of the assay. Many of the new antibody assays have been rushed to market without the necessary vetting and validation typically required of a clinical diagnostic assay. Since many of our annual colds are caused by Coronaviruses, we may have antibodies to portions of the SARS CoV-2 that are not unique to that virus. The RT-PCR used to determine the presence of the virus relies on identifying a portion of the RNA that is unique to this virus. That is pretty straight-forward. For the antibody assay to be that specific, it would need to identify a protein from the virus that is also unique and not shared with other Coronaviruses. That protein then has to be purified, either from the virus or a clone that expresses the protein and that is much harder to do. The possibility of accidentally including a non-unique portion (epitope) are pretty high. In that case we would see a lot of false positives. Only time will tell if these early results pan out.
I agree this is an excellent article.
On the issue of specificity, the New York results provide real reassurance. There is a huge regional variation in the rate of positives ranging from 25% in New York City to around 3 % upstate. Were the test non-specific, one would expect that the rates of positivity would be much closer between upstate and downstate, since coronavirus colds are pretty widely distributed. Since there are certainly some coronavirus cases upstate, particularly in the more urban areas, the false positives would appear to be somewhere below the 3.5% measured upstate.
As far as the death rate vs flu, the authors points are well taken. Yes, the death rate per infection is lower than the case fatality rate of 5 % or so, but it still comes out to around 0.8%,which is much higher than flu and, as the author points out, we aren’t done yet. People infected a month ago are still dying. Also, death is not the only bad outcome. Some survivors seem to suffer permanent damage to lungs and also strokes and that is worrisome and not seen very often with flu.
Very informative article and thanks but I continue to see reports such as this indicating that the specificity of test kits from most manufacturers remains questionable. I don’t see how it is possible to craft a productive research regime much less a contact trace and containment policy without 100% test accuracy (or very close to it).
NB: Let’s Get Real About Coronavirus Tests: There aren’t enough. Many are shoddy. Most aren’t even designed to tell us what we really want to know.
The caution about using serology as an “immunity passport” is way understated in importance at this stage in COVID serology.
The only thing the current serology can possibly tell us is that a person has or hasn’t been exposed to COVID in a way that provokes the production of some set of antibodies distinct from the set produced by exposure to other pathogens. Exposure at that level is expected to result in the body producing many different antibodies. Many of these are not protective at all, or minimally protective. Many of even the antibodies that confer some protection are not persistently produced.
We won’t know which, if any, of the antibodies that COVID exposure causes the body (of some people, there often is no universal response) are either protective or enduring until we can compare which set of antibodies a person has to their clinical history. There may well not be any protective antibodies at all, or it may be that some people can produce such protective antibodies, but others cannot. Just because a person has recovered from an infection does not at all prove that their humoral immunity must have kicked in to save them, and that therefore the set of antibodies to COVID in their sera must be protective. Non-specific and non-humoral immune factors can often fight off a pathogen without significant help from humoral immunity. Even if there are protective antibodies, there is no guarantee how long whatever level of protection they confer will last, because we won’t know how long they will be produced until we observe patients of known serology for however many months or years we wish to confirm that the protection will last.
Any talk at this stage of serology providing any sort of “immunity passport” is premature. This pathogen may not trigger the production of the antibodies we would like it to produce, for as long as we would like them to be produced, and we won’t know the answer to that question without waiting to see whether people with a given set of antibodies can still catch or transmit the disease.
The same consideration applies to herd immunity. The current state of knowledge does not allow us to assume that positivity for any COVID antibody means that a person cannot transmit COVID ever again. A population with even a 90% prevalence of non-protective antibodies would not confer any herd immunity at all. We don’t yet know the protective from the non-protective antibodies, and won’t know until we further observe the natural history of this disease as it correlates with the set of antibodies different people have or lack…
Ditto for the prospects of a vaccine. Unless we can identify which, if any, antibodies are protective and enduring, we will have no vaccine that will confer enduring protection. Most of that year and a half they talk about that we need to wait for a vaccine is the time needed to see which, if any, antibodies will prove to be effective and enduring.
We’re stuck with the name “immune system” at this point, but if we had to name the thing today, when we understand its workings (and failure to work) much better than a century ago, we would probably call it the “mitigation system” or something similar, some name that avoids overpromising categorical protection, or immunity. Immunity is the result we notice under ideal conditions, a situation in which initial exposure to a pathogen generates the enduring production of antibodies so effective that they squush any reinvasion by the same pathogen so early in the exposure and so completely, that we aren’t even aware that we’re under attack, and so early and completely that we don’t shed any of the pathogen and put others at risk either. Let’s hope we get that ideal result with this thing, but it is far from guaranteed. Perfect success is actually not terribly likely, but it is also unlikely that the humoral branch of our “immune” system will entirely fail to do us any good at all with COVID. The most likely outcome is that we will get some mitigation, but there is no way to predict exactly what sort and how much mitigation before we let months and years of observation correlate particular antibodies in the serology to particular clinical outcomes.
Not to mention bankruptcy due to medical bills.
Studies like these will get NO support from the government… they will make tRump’s numbers look higher
These articles by ProPublica are some of the best articles on the site. This one is another excellent example.
There may be other (easier) ways to extrapolate the community spread of COVID-19 in cities, communities, and regions: wastewater testing for shed viral particles. I’m kind of surprised that this isn’t getting more traction. Wastewater testing has been used as marker for opiod use, for polio outbreaks, and is in use now in the Netherlands – in fact, particles from the SARS-CoV-2 virus were found and reported in a town’s wastewater before anyone there was diagnosed with COVID-19.
Here’s a good introduction to the work being done on this.
I have to say I have thoroughly enjoyed this series of discussions. We now know just how many Infectious Disease Geeks (self included) read TPM.