Contents
As I write this, the Delta variant (a.k.a. the India variant, or B.1.617) is running rampant in the US. Despite the widespread availability of vaccinations, we have not reached herd immunity, and from the current look of things, I am seriously doubting we ever will. I am now assuming that COVID will be with us for the long haul.
Delta & The Vaccines
Delta appears to be substantially worse than the original COVID-19 variant. It's estimated to have an R0 somewhere between 6 and 9, and a higher, but still very uncertian, IFR. That means it spreads as easily as chickenpox, but isn't quite as bad as measles. Given what we know, reasonable estimates for vaccination rates required to reach herd immunity as north of 90% of the entire population. Not the adult population; the entire population.
The best data we have seems to indicate that the RNA based vaccines are moderately less effective against Delta than the previous variants. Thankfully, they appear to still be strongly protective against severe cases, and outcomes such as hospitalization and death, but depending on which data source you look at, the protection against infection may be under 50%. The J&J vaccine appears even worse. Thankfully, trials of an extra Pfizer dose as a follow up to the J&J were recently completed (though I can't now find a link), and the results were encouraging. At this point, I'd strongly recommend anyone who got a J&J first dose to locate an MRNA second dose.
There's also solid evidence in support of a third RNA dose for high risk individuals. At the moment, I'm not personally rushing out to locate a third shot, but if I was in a high risk group, I'd be seriously thinking about it. As of Friday, the CDC is recommending a third shot for high risk groups.
Vaccination
First, if you are reading this and are unvaccinated, stop and go get vaccinated now. The only reason not to would be if you have been told personally by your doctor not to. You are putting yourself at risk, but honestly, I don't really care about that. If you were only playing russian roulet with your own life, I'd not really care. However, by not being vaccinated, you are also now putting the lives of everyone around you at risk as well. That is simply irresponsible.
As of today, only something around 50% of the US population is vaccinated against COVID. However, there are some serious caveats to even that number.
- There is immense variation between states and counties. This is pretty wildly covered, so I won't waste space here.
- Kids under 12 are still not approved by the FDA. At this point, I'm starting to seriously wonder about the delay. What the heck is taking so long? Given how much more several Delta appears to be for kids, this delay is extremely problematic.
- Those rate include both MRNA and J&J vaccines. Given the new data on J&J effectiveness against Delta, that's starting to seem like a problem.
Regardless, our overall vaccination rates are way too low. And currently show no signs of increasing quickly enough to matter for near to mid-term planning purposes.
Covid is Endemic and Likely to Remain So
Based on what we're seeing now with Delta, it looks very unlikely that COVID will ever be eliminated from the US population. The required vaccination rates to have a chance of suppression appear impractical. Worse, the virus continues to mutate. We now have solid evidence that the virus can evolve resistance to vaccines, and wishing otherwise is looking increasingly futile.
On the personal front, I'm starting to live my life as if this is our new normal and likely to stay that way. Thankfully, the MRNA vaccines do appear individual protective against severe outcomes. I am personally returning to masking everywhere I go, and doing my best to avoid crowded places. However, based on the reasoning that things are unlikely to get much better, I am not retreating from life. I am traveling. I am eating out. I am simply doing so while taking every reasonable mitigation tactic I can. I am socializing with friends who have been vaccinated. (I will not knowingly socialize with any unvaccinated adult. Mostly because of the implied stupidity and lack of care for my health involved.)
On a public policy note, I'm now going to do something I haven't done in these writeups previously and dive right into the "politics". I can't tell you how utterly disgusted I am that public health measures have become a political litmus tests.
We need to drive vaccination rates. The carrot has been tried, it's time for the stick. I fully support vaccine mandates. My general rule of thumb is that any place which can require a measles vaccine (e.g. school, most workplaces, etc..) should require a COVID vaccine. The common objection here is "freedom!" - which quite bluntly is utter bullshit. Freedom of choice does not mean freedom from consequence. You should be free to decline vaccination, that I completely agree with. You should also be free to be unemployed, required to home school your kids, and unable to participate in public life. Choices have consequences; deal with them. Freedom is not free.
On the matter of public schools, they must be kept open - at almost any cost. We can and should take every practically effective mitigation measure - vaccine mandates for staff and kids over 12, masks, social distancing, improved airflow, etc.. At the end of the day though, keeping public schools open should be a core goal. The cost to both our kids and society as a whole of another lost year - because let's be honest, online education last year was a disaster and most kids got nothing out of it - is unacceptable. If you care about equity at all, getting schools open so that parents can work should be top of your list.
To be very explicit about the combination of the two previous points, any parent not willing to comply with school safety measures - masks and vaccines, etc.. - should be required to home school their kids. We should provide online programs where practical to make this easier, but the responsibility and cost should lye with the families choosing not to participate. Let's be clear, it is a choice. If a family which chooses not to comply with requirements for public schooling can't provide an alternate education framework, well, under well settled law that's child neglect. We should deal with it as such.
In practice, people will lie about vaccination status if we adopt the policies I've just advocated. How should we deal with that? One school of thought is to make it harder to lie (e.g. government issued vaccine passports, etc..). There are some reasonable steps we can take here, but we run into some obvious data privacy and constitutionality issues pretty quickly.
Instead, I propose a simple alternative - liability. An individual who lies about vaccination status should be fully liable for their conduct, both in matters of civil law and criminal law. If your actions kill someone, that's manslaughter. If you're actions put someone else in the hospital, their insurance should sue you for the costs of care, and they should be able to sue for monetary damage in line with any long term effects. This is in line with our legal structure, so much so we may not even need any specific law to enable it.
In the US, we've been hovering around 60k new cases a day for the last month. That's a heck of a lot better than I expected to be seeing right now. We've recently seen the beginning of what might be an uptick, and in particular an increasing fraction of cases coming for new variants, but so far, nothing I consider too worry worthy.
On the vaccine side, we've administed roughly 140 miilion doses to date, and are up to around 2.7 million doses per day. (Bloomberg) Given a month ago, we'd only administed 60 million doses and were running at 1.5 million per day, these are awesome numbers. There's still a long way to go, but the progress to date has been impressive.
For context, we'll need to have administered something around 460 million does to reach 70% coverage. Given it seems even a single dose of the existing vaccine provides strong protection, I expect we'll see dramatic improvements in case counts before then. If we guess it takes 300 million does to get 240 million first doses, then we can reasonably expect to be there in roughly 2-3 months from today.
There's still need for concern with the new variants as there's potentially room for a fourth wave of infections to spike between now and then, but overall, I'm seeing a much more hopeful picture than I expected to be. Right now, we seem to be nearing the end of this thing, and the window for serious uptick in cases seems to be closing rapidly. If we can get through another few weeks without a huge surge in cases, we'll be in pretty good shape.
And, even if we did see a sharp spike in cases, given the high vaccination rates in older adults we'd expect death rates to be lower. So, we might get out of this thing with "only" 600-700 thousand dead. "Only".
Minor update today.
Per the CDC, we're up to about 65 million vaccine doses administered. We're still hovering just under 1.5 million doses per day. There's reason to hope this is about to improve, but it hasn't happened yet.
The good news is that cases are down. Way down. As of yesterday, we had 59 thousand cases reported. The last time we were that low was mid October. This is hugely good news, and I will freely admit that I have been shocked by how fast the case numbers are dropping. From what I can tell, no one really knows why this is happening, so I'll refrain from joining the speculation. Death rates are also dropping, though more slowly (as to be expected given inherent lag).
Looking forward, I'm still concerned about the new variants on the horizon. Things are looking much better now than I expected them to be, so I'm starting to hope the UK variant triggered wave won't be too awful. We're getting more and more evidence the vaccines at least prevent severe cases and death with the new variants; that's awesome news! On the other hand, vaccination rates are still low and at current rates, the variants have several months to get established and cause headaches. We desperately need to increase vaccination rates. There's some hopeful signs - e.g. the availability of the J&J vaccine in the near future - but I won't fully believe it until it happens.
As mentioned last time, I've been lightly following developments with the new variants and new vaccines. Unfortuntately, what I'm seeing is looking more concerning not less.
The biggest item is that the South African variant appears to widely reinfect those who have already had the original strain. (Washington Post, twitter thread) This is seriously bad news as it means that existing infections to date don't contribute to slowing the spread of this new strain. The good news is that vaccines do appear effective against this strain, but at lower rates. We're still figuring out what those rates are, but they appear to be substaintial drops from the high 90s we saw for the first strain.
And this is on top of the fact that the UK variant appears to be becoming broadly established around the world. That one may have a higher mortality rate than the original, though that's not conclusively confirmed. Importantly, it does appear to spread 50% more easily. Thankfully, vaccines appear fully effective here.
On the good news side of things, manufactures and the FDA both seem to be getting set up to rapidly iterate on vaccine boosters for the new strains. I've seen some takes which are very positive on how fast this can be done, but frankly, I don't know enough to assess how realistic that is.
As of today, the US has administed something along the lines of 39 million doses of the vaccine, and we're running about 1.3 million doses per day. If we assume 65% of the population needs vaccinated to reach herd immunity to the new strains, then we've completed about 9% of the task, and are increasing that by about 0.3% per day.
To be honest, that's probably a best case. Given the lower efficacy numbers we're seeing with the new vaccines, I'm guessing we'll need higher vaccination rates too. Much higher.
Even taking the best case, the situation is not looking good. At current vaccination rates, it'll take almost 11 months to achieve the required level of vaccinations. There is an optimistic scenario here, but it requires sharp increases in the daily vaccination rates and even then, we're looking at June/July at earliest.
At the moment, it's looking more and more like we're going to be facing (at least) two waves of the pandemic.
The first is the one we're in. It's fairly far along, we're approaching useful herd immunity levels quickly, and we could see the light at the end of the tunnel. The UK variant is about to bite hard, but we're in decent shape overall.
The second is the one caused by the South African variant - or another variant we haven't seen yet - for which reinfection occurs at high rates. This one makes us much more reliant on vaccinations. As mentioned above, that extends the timeline and gets us into some uncharted territory as we don't know how protective the vaccines will be.
I'm feeling very reminded of a year ago this time, when COVID was just getting established, we didn't know the full story yet, people hadn't yet panicked, and a potential life changing event was on the horizon. At a minimum, I strongly suspect we're all going to be taking our annual covid booster shot for the forseable future. Whether that's all that is required, or whether more profound changes to our definition of normal are coming, I'm utterly unclear.
As of today, the US has had 25 million confirmed cases, 417 thousand confirmed deaths, and 20.5 million doses of vaccine administered. I'm so glad to be able to finally include that last number.
If we assume for our back of the envelope math that each dose works out to one half of an immune person (either because they got two or because each person is less prone to catch it even with one), and that the 5x estimate of total cases versus confirmed cases still applies, that means we have something like 135 million people now immune. This is already ~40% of the US population.
Given our current infection rates of around 175k confirmed cases per day, and vaccinations of around one million doses per day, that gives us around 1.375 million newly immune people per day. That means we should hit 65% of the population in roughly 58 days, 70% in 70 days, 75% in 82 days, etc...
It is important to highlight that this back of the envelope math isn't exact by any measure.
- As we get higher and higher immune rates, we should expect to see transmission rates start to drop. On the other hand, with the new UK variant having a higher base tranmission rate, that effect may be muted or reversed. Given that, the linear projections I used above are probably inaccurate, but in which direction I make no predictions.
- The 5x ratio of confirmed cases to total cases is based on estimates from early in this pandemic. It's probably wrong. By how much, I don't know. The fact I can't find more recent refined estimates for this is a sign of just how badly managed our state and national response has been. We do at least have some data from the CDC from back in November which seems to indicate the estimates aren't too far off.
The only real takeaway is that we're within months of the end of this. We don't know exactly how many months, but it's much more likely to be 3 than 12 at this point.
I'm loosly tracking the emerging news about vaccine variants. So far, nothing I've seen materially impacts the picture just described. As previously mentioned, the UK variant may increase transmission in the near term - though probably not severity or death rates. I'm more concerned by the news that the Brazil variant may be leading to a large number of reinfections. I consider the evidence of this poor to date, but if true, that could be concerning long term. We know that the UK variant doesn't impact the effectiveness of our current vaccines; there's reason to hope for the same with the others, but we just don't know yet.
It's been a week of mixed news.
On one hand, the Moderna vaccine has been approved by the FDA. Vaccination campaigns have begun with many front line medical staff already being vaccinated (with the Pfizer vaccine to date). That's wonderful news.
On the other, we have confirmed the virus is mutating and there are some reports that strain spreading in the UK and Europe spreads easier. (I can't find scientific data to back that yet.) There's no evidence that the mutated strain is more lethal or resistent to the vaccines.
I don't think this really changes anything in the near term. It does make it more likely that COVID will be with us for a while, and that we may need to evolve the vaccine as the virus evolves. (Like influenza and the flu vaccines each year.) This is one of the semi expected outcomes of having failed to tightly control transmission and now having so many people infected. (The more opportunities for mutation, the more likely we were to see one.) At the moment, I'm not worried by this news, but it is something to keep an eye on.
We've also seen the first of the expected operational confusion impacting vaccine distribution.
I'd gotten a key detail wrong in my Dec 12th comments, please see the correction below. The vaccine section is essentially entirely rewritten, and the tone of the forward looking section got revised to reflect more available vaccine doses.
As of today, the US has had 16.2 million confirmed cases, and 298 thousand confirmed deaths. This is over 65k deaths in the last 5 1/2 weeks alone, and we're rapidly approaching (w/in the next week or so) the point where 1 in 1000 people in the US will have died of COVID-19. Assuming a roughly 5x case to confirmed case ratio, we've had around 81 million cases in the US to date.
I'm expecting these rates to continue increasing over the next few weeks. Without the availability of vaccines, we'd likely be up around 100k deaths per month by early Spring 2021. Now that we have some availability of vaccines (see below), it'll be a race to see how many people we can vaccinate vs how quickly people are infected in the general population.
Correction: I'd originally confused the Moderna vaccine and the AstraZeneca one. As a result, my original take here was flat out wrong and has been updated 12/14/2020.
As of today, we have two vaccines which are either approved for use in the US, or very nearly so. I'm absolutely thrilled by this development. As I stated a few weeks ago, I was not expecting vaccines to be available this soon.
The Pfizer vaccine appears to be very effective. The reported efficacy is 95%, but I've seen several folks point out that the result is essentially statistically indistinguishable from 100% efficacy. The single "severe" case was borderline, and the stats pivot on that singe case. The challenge with this vaccine is a) the need for a cold chain distribution, and b) the limited supply of 100 million doses (enough for 50 million people) in the near to moderate term. The supply side is infuriating, as apparently the US government declined to purchase an option on 100 million more doses. WTF.
The Moderna vaccine also appears to be have high efficacy, and distribution appears more straight forward as this one doesn't require anywhere near as extensive cooling, which is a huge win. At the moment, it looks like we're about a week out from approval on this one as well.
Over the next few months, logistics of vaccine distribution is going to be a major challenge. I'm expecting lots of fuck ups, both because the outgoing administration has proven themselves very good at those, and because this is simply a hard problem.
Let's use some back of the envelope math to get a feel for what the next few months could look like. We're seeing around 200k cases per day today. For both scenarios we'll assume a 5x case to confirmed case ratio, and we use the estimated 80 million cases to date from above. That means we have roughly 1 million new infections per day.
- Pessimist Scenario - If we did nothing (and ignored the non-linear nature of disease progression modeling), we'd reach 100% of the population roughly 8 months from now. That won't happen for all kinds of reasons, but it gives us a feel for what to expect at the outside.
- Optimistic Scenario - If we assume 65% immunity is enough to dramatically slow infections due to herd immunity, we'd be looking at 4.5 months at current rates. That's only 135 million people remaining to herd immunity, and we're looking at the availability of 100 million persons vaccinated based on current supply. We could be down to as little time as it takes us to deploy the vaccines. (Current infection rates are almost certain to provide the other 35 million cases in any reasonable deployment timeframe.)
Both of these are rough estimates a best, but they give a flavor for what we might be looking at. I'm reasonable confident in saying the worst of this pandemic will be over by summer, the only question is how bad it'll be between now and then.
Folks are starting to put out more sophisticated estimates for a path to herd immunity based on vaccination campaigns which are useful for exploring the possibilities.
I am fully expecting the next 2-3 months to be ugly. At this point, there is a light at the end of the tunnel, but we're still a ways from the end of this. If we reacted well and executed well, we could minimize the costs between now and then, but frankly, the last year has made me anything but hopeful we will.
As of today, the US has had 9.47 million confirmed cases, and 233 thousand confirmed deaths. That puts us just under 30 thousand deaths a month. This is right in line with my projections from back in July, and if anything, maybe a bit lower than I expected.
I can't really find much in the way of updated population infection estimates (e.g. antibody studies, prevalence studies). What I can find is from back in July (i.e. quite a bit lagged), with two major sources:
- The CDC's estimates still have total case counts somewhere in the 4-8x range of confirmed cases.
- A Stanford study came up with a roughly 8% estimate. Using confirmed case counts from end of July, that's at least a 5x multiple.
If we assume those ratios still hold - and given the sky high test positivity rates we've been seeing across the country, if anything, they may now be too low - I think it's safe to say we've had something in the order of 50 million cases in the US so far. Given the number of confirmed deaths, that puts our IFR somewhere around 0.5%.
At the moment, there's serious concern that cases are broadly trending up across much of the country and that selected areas are reaching capacity limits for local health facilities. At least at the moment, I don't see anything to make me thing we're going to enter wide spread care failures; we'll probably see isolated incidents, but nothing broad spread.
On the good news side of things, we have much improved standards of care since this things started. Survival rates are up - though importantly, not "way up". Between basic care, availability of cheap and effective steroids, and the first wave of antibody treatments hitting the market, thinks are looking up on the care front.
I still remain skeptical of a vaccine in the immediate near term - and recent developments have supported my skepticism - but it's looking more and more likely that we will have a vaccine within the next year or so. My personal guess is that we'll start seeing availability sometime next summer.
I expect that a vaccine is not going to be a miracle cure. Between the likelihood that initial vaccines are likely to only be 50-70% effective, and the strong vaccine hesitancy which exists in this country, I will be suprised if availability of the vaccine does anything more than slow the spread.
At current rates, we're seeing about 2.5 million confirmed cases (and thus likely around 12 million total cases) per month. We're at around a 15% population infection rate today, and increasing by about 4% per month. Projecting that out, we'd expect to start hitting leaves of practical heard immunity (60-70%) late next year.
If we get a vaccine, or case counts tick up further, that date may pull in some. If we start seeing dropping transmission rates - entirely possible as even partial herd immunity effects likelihood of any individual infecting another - we may see case counts drop slightly and time lines extend. During that time, we're looking at around another 700 thousand deaths.
Overall, I see a lot less uncertainty in the progress of this pandemic than I did a few months ago. It's going to by ugly - though as I've said before, not catastrophic. My personal projection is that COVID will become our new normal over the quarters ahead. Life will go on, many people will get sick, some will die. But overall, life will go on.
As of today, the United States has had 6.95 million confirmed COVID-19 cases, and 202 thousand deaths. That's a plenty grim milestone. It's also worth noting that the US is now worse on a per capita death rate has now passed Sweden, and we currently rank 11th worst in the world on this metric. (Well, out of those countries which report at least, and there's a number that don't.)
Looking ahead a bit, the city of Manaus, Brazil may be showing us <https://www.technologyreview.com/2020/09/22/1008709/brazil-manaus-covid-coronavirus-herd-immunity-pandemic/>_ what things would look like if this continues to burn through the population. There's reason to believe that they have actually reached (or at least nearly reached) herd immunity. In the process, roughly 1 in 500 people have died. This is about 3x worse than where the US stands today. The estimated IFR is around 0.3%.
If that matched our experience, we'd expect to see somewhere around 600-800k deaths here in the US. However, even assuming Manaus actually has reached herd immunity, it's important to note the US experience may be much worse. Manaus is a fairly young city overall with less than 6% of the population over 60. That same number for the US is roughly 20%. Given we know the risk on this increases greatly with age, that gives us strong reason to suspect the death rate would be higher here.
I remain interested in trying to understand such worse case scenarios as I remain very skeptical that we're going to have an effective vaccine any time soon. The Russians have been caught manufacturing data on their vaccine and the US vaccine studies are coming under seriously problematic political pressure. I remain skeptical that we're going to have an effective vaccine in wide deployment any time in the next six months, and maybe not for a full year or more.
If projections above (and similar ones I've given before based on estimated IFRs) turn out to be right, we're looking at increasing the death rate for the year by ~25%. That's bad, but it's also not catastrophic. Nearly 3 million people die each and every year. Most years, we - as a society - never really notice unless one of those deaths strikes close to home.
As of today, the USA has had 6.08 million confirmed cases, and 184 thousand confirmedc cases. Subtracting out the counts from two weeks ago, that's 700 thousand new cases, and 15 thousand new deaths. That gives us roughly 1.3m new cases in the month of August and 27k confirmed deaths for the same.
For context, if we assume the 3-10x estimate for number of actual cases vs confirmed cases still holds, this would imply that there has been somewhere around 20 and 60 million cases in the US to date. That's potentially as much as 20% of the population. There's reason to be a bit skeptical of that since the most recent antibody study results I've seen aren't anywhere near that high, but something in the 5% range seems fairly plausible. So, "only", 1 in 20 people in the US.
The other bit of context is that the raw CFR over the last month has been around 2% and the cummulative CFR for the whole period has trended down to ~3%. As before, if we assume only some fraction of cases are getting caught, that puts IFR somewhere in the 0.2-1% range. There are much fancier estimations out there; I find doing back of the envelope numbers like this to be useful in keeping context though.
The major thing I'm tracking virus wise is that we've now had our first two confirmed cases of reinfection. The fact we're seeing cases isn't suprising; with over 6 million cases in the US alone, it would be surprisig if we didn't! It's clear from the two cases which happen to be documented, and our relatively poor testing situation, that there are probably many more cases out there. On the other hand, this doesn't seem to be widespread just yet. The real question is what the population level immunity drop off looks like; that'll be a big factor in reducing spread rate w/ or w/o a vaccine. So far, I don't see any strong reason to worry just yet.
As of today, the United States has had 5.37 million confirmed cases, and 169 thousand deaths. That's roughly 600k new cases, and 12k deaths in the last two weeks. This continues to be right in line with my projections from July 19th.
There are some tentative signs of good news developing in the last two weeks.
First, SalivaDirect, a new inexpensive COVID saliva based test developed by Yale on a not for profit basis, has received emergency approval by the FDA. This is a really big deal as the supplies to run the test cost under $3 and the testing protocol is public and could conceivable be scaled very widely. If we can get to the point where it's reasonable to test everyone - and I mean everyone - every couple of days, we can realistically control this thing and go back to a much more usual existance.
Second, a monoclonal antibody therapy from Eli Lilly entered third stage testing. Initial results look promising, and the science behind such an approach is fairly well understood. If we can identify a treatment which meaningfully decreases mortaility and scale it appropriately, that could be a game changer. One concerning sign is that I haven't seen much on plans to scale production; Operation Warp Speed which is funding vaccine production efforts doesn't seem to be funding treatments. I hope I've simply missed a headline here.
Third, Sweden - which has been following a fairly open strategy all along - appears to be seeing both new cases and death rates drop sharply in recent weeks. This is encouraging as it gives us an idea of what a mostly out of control spread scenario looks like, and it's lot less bad than it could have been. Now, Sweden's total per capital death rate is still a lot higher than it's neighboors and it's population has a lot lower risk profile than the United States, but still, this is encouraging to see. Do note that just because Sweden hasn't had formal shutdowns doesn't mean that individual behavior hasn't radically changed; I'd read this more as a hint as to what spread looks like in a health population practicing social distancing measures than anything else.
Fourth, preliminary reports on a treatment called RLF-100 appear to be very promising. It's important to note that these are very early results, and I haven't see enough on this to know how real this is just yet. If it works out as a treatment for several COVID cases, this could again be a game changer. Interestingly, production would not need to be scaled anywhere near as much as a vaccine since you only need to treat the folks who develop severe cases. In the United States, that would mean 50-100k doses a month at current rates.
This will be a very short update as nothing major has changed in the last two weeks. The USA is at 157k deaths with 4.71 million confirmed cases. That's 13k confirmed deaths in the last two weeks, which unfortunately puts us right on track from my projections last time. As before, there's some evidence that CFR is trending slightly lower, but the data is noisy enough to be hard to interpret.
I'd really like to see someone perform an analysis of CFRs for individual {two week periods x locales} and then plot the computed estimates against test positivity rate. I suspect from the data I've looked at that the national CFRs are being driven quite a bit higher than reality by high positivity rates (and thus low estimations of total confirmed cases). Unfortunately, no data I've seen would imply an IFR much below 1/2% at the absolute best.
I'll close by pointing to a nicely written article by fivethirtyeight which has the best description of the calculated risk decisions each of us are making day by day I've seen so far. The only thing I fault the article for is failing to acknowledge that this is the same decision procedure we have always applied - consciously or not - all that has changed is the (estimated) risks.
The big question being discussed this week was whether COVID-19 provides any form of extended immunity. Such immunity is a key part of any herd immunity strategy - whether infection or vaccine based. The best description I've seen so far is from ArsTechnica. The summary appears to be "it's complicated", but there's no particular reason to panic just yet.
One weirdly positive bit of news buried in the discussion of antibodies vs t-cell immunity is that our current surveillance testing only detects antibodies. If - and this is a big if - it turns out than many people loose antibodies quickly, but retain at least some partial immunity via other mechanisms (t-cells?), then our estimates of the number of people infected so far may turn out to be low. That would be good news for IFR if true. I want to emphasize that we just don't know, and shouldn't place much hope in this.
One apparent bit of very good bit of news, buried in all the bad news, is that death rates definitely appear to be trending down. As of today, there have been 143k deaths out of 3.83m confirmed cases. This a CFR under 4%.
If we look at only the cases and deaths since June 12th, we've got 27k additional deaths and 1.73m additional confirmed cases. That would give a lower bound on CFR of around 1.5%. It's a lower bound as deaths are a lagging indicator, and it's hard to say how much the additional death number would increase from currently active cases.
If we take the deaths as of today and the cases as of July 2nd (to try to adjust for the lag in deaths), we'd be looking at 27k additional deaths and 640k additional cases. That would have our CFR back at something around 4%.
Conclusion? It's really too early to say what's going on with CFR. It might actual be trending down, or we might be fooling ourselves by combining metrics with different lags. It's impossible to say.
I don't want to be alarmist, but the current situation in the USA is distincly "not good". We appear to be following a path of barely controlled burn through. As bad as things currently are, the fact we're seeing shutdowns again mean things aren't fully uncontrolled either. For reference, fully uncontrolled burn through screnarios are the ones which completely swamp hospital capacity and we see CFRs north of 20%. We're not seeing that, and I doubt we will for any sustained period.
My current personal best guess is that IFR will end up someone around 1/5th of the current estimated CFR. (So, around 1%.) I expect we'll continue to see US states relax and then tighten restrictions with the effect of keeping R somewhere close to 1. Given this, I am expecting to see a slowly increasing number of deaths for each month until we have an effective vaccine. As a ballpark, let's say around 20k increasing up to around 50k per month, or around 150-300k over the next 6 months. At some point we'll start seeing R drop due to partial herd immunity, but practically, I suspect we're going to be hovering around R~=1 for the forseable future.
I really hope I'm wrong; these are pretty terrible numbers. But on the other hand, it is important to keep perspective. Somewhere around 2.4m people died (of all causes) in 2019. If we project 600k from COVID, 2020/2021's death rates will definitely be well above average, but they're not going to catestrophic either.
Just a collection of links for the moment.
WSJ, For Struggling Small Businesses, Bankruptcy Law Change Comes Just in Time
CNN, Covid-19 immunity from antibodies may last only months, UK study suggests
As of today, the United States has had 130 thousand deaths out of 2.74 million confirmed cases. This gives us an estimated CFR of ~5%, which is in line with the 6% estimate from a few weeks ago.
This week, the CDC reported results from antibody studies which seemed to show actual case rates were more than 10x higher than confirmed cases. I'd honestly love to believe this is true, because if it is, it means the IFR is somewhere around 0.5%. However, I think there are some reasons to be cautious here.
- First, and I hate saying this, the CDC has come under a lot of political pressure. That may be biasing the results.
- Second, the absolute infection rates in most of the regions studied is low. From the linked to paper, the false positive rate on the test used was just under 1%. That would seem to put the results out of the range of likely error, but it does mean the claimed ratios are potentially too high. In particular, the highest claimed ratios appear to be from the lowest absolute percentages (and thus most influenced by false positives.)
- Third, and this is the biggest one, the data is old. The most recent reported result is from May 2nd. For a result published almost 60 days later, that is flat out suspicious.
Putting it all together, I'd be willing to say that case rates are at least 4-5x higher than confirmed via testing based on these results, but I wouldn't go beyond that. (As much as I'd like to.)
A couple weeks back, we learned that dexamethasone, a common steroid, appears to reduce death rates in severly ill covid patients by about 20%. This is wonderful news, both because it would reduce our observed CFR, and also because this is a generic medication which is already widely available and cheap (less than $8 per dose). That is by far the best news we've gotten to date.
This week, we're seeing efforts to scale the collection and distribution of blood plasma from recovered covid patients. As mentioned previously, we have good reason to believe that such a strategy works, and can help reduce the severity for many patients.
Putting these two together, that's a dang good bit of news. I expect we'll start seeing the CFR trending downward over the next few months. There's some hope we're already seeing that in the national data, but there's also a bunch of other interpretations possible there.
I will note that I remain sceptical of the possibility of a widely deployed vaccine within the next 12 months. I suspect we will see one, but almost certainly not this year, and next year is a merely a hope. In theory, timelines could be accelerated with good planning and coordination, but we haven't exactly seen much evidence of that recently.
On the topic of antibody studies, we do have one small update from NY State `in minority cummuniy churches <https://www.governor.ny.gov/news/amid-ongoing-covid-19-pandemic-governor-cuomo-announces-results-states-antibody-testing-survey>'_. I am increasing nervous at the fact the state of NY has not been publishing updates to their antibody study.
Despite the relatively scarcity of new data, it seems like there is an emerging consensus that the infection fatility rate for COVID-19 is somewhere slightly under 1%. The case fatality rate on the other hand seems to be hovering right around 6% for all of the data sets we have. At the national level, we currently have 2.1 million confirmed cases, and 116 thousand deaths for a CFR of 5.5%. As discussed previously, deaths are skewed very strongly towards the elderly, so what these numbers look like in each community is strongly dependent on demographics, but the rough numbers give us a rough idea of what we're looking at.
One correction to the writeup below. The study I referenced on hydroxychloroquine has been heavily critized and retracted. Other studies are still supporting a fairly skeptical attitude here, but the study which initially appeared fairly conclusive turned out not to be.
The number of deaths per confirmed case is disturbingly high. The NYC numbers [1] as of today are 195,452 cases, with 16,469 confirmed deaths and another 4,747 probable. This works out to a more than 10% death rate, concentrated almost entirely in older adults [2].
Thankfully, there's a big difference between confirmed cases and number of people infected. The best evidence we have to date is the new york antibody study [3] found 24.7% of the population to be positive for antibodies implying they had been previously infected. With a population of 8.6 million that would mean actual case counts were around 2.1 million, ad that the death rate is actually closer 1%. It does make me nervous that the last update on these numbers I can find is now three weeks old though.
There is no evidence for reinfection at this time. There were some initial reports from South Korea of potential reinfection cases, but those have now been thoroughly disproven. The cases in question were either false positive on tests, or individuals shedding dead virus. From other viruses in the same family, we have every reason to expect a prolonged immutity period of at least a couple of years. Neither point is confirmed yet, but we can be reasonable confident that if there wasn't a substaintial period of at least partial immunity that we'd have seen that by now.
There is some evidence of lasting effects even in younger people. However, all of the cases reported so far are in very small absolute numbers. That might change, but at the moment, we have no reason to believe that any large fraction of the population has long term complications following recovery.
I have focused on the NYC data - mostly because it's the largest sample size with the fewest known bias problems - but the same general picture appears everywhere else we have data as well.
One key statement is that for most of the US, containment has failed and is no longer a viable strategy. This is definitely true in NYC; there's no possible way to contact trace 100s of thousands of cases. This is not true for many other areas of the country which have much lower case counts which is one legitimate reason that responses will and should differ in different locations.
Given that, we're basically looking at having to let this burn through the general population. The only good news is that a) the death rate seems to be about 1%, b) it appears to be heavily concentrated in older adults, and c) at least in NYC we appear to be at least a fourth of the way there. Putting that in perspective, roughly 0.8% of the population dies from natural causes each year. Given that, we're talking about an effective doubling of the annual death rate. That's horrible, but it's also nowhere near a worst case scenario.
We strongly suspect that plasma treatments work [4]. They're hard to scale, but we have every reason to believe from history that the approach is workable and we have a number of studies which confirm this.
We know that remdesivir shortens recovery times [5]. It may also have a small effect on mortality, but that's unclear. The important part is that by shortening recovery times by roughly 30%, our hospital capacity is effective increased by 40%. That's huge because it helps us be a lot more confident we can avoid the hospital overload scenarios which could drive the death rates through the roof.
Despite what certain idiots might tell you, we know that hydroxychloroquine does not help [6] and actually appears to harm. There's still room for further evidence here changing the picture, but at the moment, it looks like taking any of the drugs in this family is a damn bad idea.
I consider the odds of having an effective vaccine widely available before this has finished burning through the general population to be quite low. I'd love to be suprised, but at the moment, I'm assuming this is a non-factor.
A few weeks ago, there were reports [7] that survival rates for patients placed on mechanical ventalators were very low. Unfortunately, the media badly misreported this study. The reality is that more than 50% of the patients in the study were still in treatment (i.e. alive at the time of publication). The scary numbers everyone (including me) saw were reporting the fraction of people who'd died out of those who'd either died or recovered at that point in time. Until we have updated numbers - which oddly, I haven't seen yet - the results could be anywhere between a 60% recovery rate and a 90% death rate. Really, we have no idea.
| [1] | https://www1.nyc.gov/site/doh/covid/covid-19-data.page |
| [2] | https://www.statista.com/statistics/1109867/coronavirus-death-rates-by-age-new-york-city/ |
| [3] | https://www.livescience.com/covid-antibody-test-results-new-york-test.html |
| [4] | https://www.nature.com/articles/d41587-020-00011-1 |
| [5] | https://arstechnica.com/science/2020/05/the-antiviral-remdesivir-shortens-covid-19-recovery-times-study-shows/ |
| [6] | https://arstechnica.com/science/2020/05/hydroxychloroquine-linked-to-increase-in-covid-19-deaths-heart-risks/ |
| [7] | https://www.bloomberg.com/news/articles/2020-04-22/almost-9-in-10-covid-19-patients-on-ventilators-died-in-study |