statistics – TJWebbMD.com

COVID-19 Questions and (attempts at) Answers, Part 3: Isn’t a surge a good thing? Herd Immunity and the RECOVERY Trial.

June 29, 2020July 18, 2020 TJ WebbLeave a comment

I’ve had quite a few questions about COVID-19 put to me by friends and family members recently, and so last week I had intended to begin trying my best to answer them. This plan had to be put on hold when Waco (and various other cities in Texas) issued a requirement to wear a face mask inside of businesses and restaurants, and the whole world sort of lost it’s collective mind. I think things have calmed down now, at least locally, and as I’ve driven to clinic and back and the one or two other places I couldn’t really avoid going, I’ve thankfully seen a noticeable increase in masking, either in compliance with this decision or in response to the efforts of so many to share reliable information on the benefits and safety of wearing a face mask. Thank you all for fighting the crazy amount of misinformation out there. For my take on wearing masks you can read my previous blog post on masking.

Now in 8-bit Color!

Now that we’ve made it through another week without another viral misinformation video, I’m finally taking the time to sit down and write that original post I had planned on. I’ve tried to limit myself to just two paragraphs for each topic (paragraph length unspecified), but given just how many questions

Due to length, I’ve broken this post up into multiple parts.

Warning: These got really ‘mathy’ on me before I realized it was happening.

Part 1: Is the rise in cases just due to more testing?

Part 2: What about antibody testing and asymptomatic cases?

Question #3: Isn’t a surge a good thing since it will give us herd immunity?

The concept of herd immunity, susceptible persons being protected from infectious diseases by a sufficiently high number of people in their community already being immune, was controversial even before the COVID-19 pandemic. I don’t mean it was a controversial area of epidemiology; the science behind it is very well established and pretty straightforward (and if you are going to read about the eradication of smallpox from that link, you should also read about a man called Onesimus, a slave in Boston whose knowledge of West African inoculation saved hundreds or thousands of lives and paved the way for Edward Jenner’s eventual invention of vaccination techniques). I mean it was something that we’ve had to argue about constantly in recent years because the anti-vaccine movement uses herd immunity as one of its many arguments against vaccination, while at the same time undercutting its effectiveness by seeking to decrease the number of people who are immune through being vaccinated. The idea is great in principal; just weather the storm now and then we will all be safe from the virus forever. The problem (one of the problems, for there are numerous) is that we don’t yet know exactly what percentage of the population needs to be immune to confer protection to everyone else. Most estimates have put this number somewhere between 60-70%, but a recent model published in Science estimates it at a much more attainable 43%. These numbers are based on several parameters that tell us both how easy the virus is to spread and whether certain activities, situations, or even individuals are more likely to spread it than others (you can read about the median reproduction value and dispersion factor if you want to dive a bit more into the math of it all). Because these numbers are incredibly hard to definitively determine in the midst of a pandemic, any percentage we arrive at is going to be a best guess; an estimate derived from multiple assumptions that will only be proved wrong if many more people get very sick even after we’ve achieved the required numbers for herd immunity. Herd immunity is a gamble because Virology, during a pandemic, is an applied science; the virus will correct all of our miscalculations and false assumptions for us. (other questions, such as whether immunity to SARS-CoV-2 is indeed long-lasting and whether the virus will mutate in such a way that it causes future outbreaks despite our acquired immunity are also important, but outside the immediate scope of the discussion).

*This is from early in the pandemic, but a great visualization tool

But even more important than the difficulty in calculating the necessary percentage of people being immune to confer protection to everyone else is the question of how dangerous it is to get there in the first place. Let’s talk about measles for one moment. We know that the herd immunity required for measles is somewhere around 93%, which is part of the reason we have seen outbreaks of the disease recently in areas that have a substantial anti-vaccine sub-culture; it isn’t hard to fall below that number. Let’s say there wasn’t a Measles vaccine; that means 93% of people would need to develop immunity by living through the disease. With modern medical advances the case fatality rate for measles is a lot lower than it used to be, but it is still around 2.2%. This means that in a country of 330 million people that had no immunity to measles, 306 million would need to contract the disease to confer herd immunity to everyone else; of those, 6.75 million would die, not to mention the longstanding residual neurological deficits and other health complications in tens of millions more. Without effective vaccination, herd immunity would simply never have been an option for Measles; the cost in human life and suffering would just be too high. But what about COVID-19? We know that SARS-CoV-2 is thankfully less contagious, and we believe less deadly (see the last post for a discussion on this) than measles, but is it enough to make herd immunity a viable option? Let’s apply those same calculations based on the current estimates we have for infection fatality rate. If we accepted a 1% death rate estimate, then to achieve the widely accepted 60% mark for herd immunity we would see 198 million cases and 2 million deaths, while if we accepted the recently released 43% estimate and assumed an even more conservative 0.5% death rate, that would be 709,500 deaths; and neither accounts for the longstanding health deficits or the cost in human suffering of those who survive, or the other deaths and suffering that come with an overwhelmed mid-surge healthcare system. Now, could we devise some clever epidemiology strategy that uses emerging data about the already-immune, super-spreaders, natural resistance, new drug therapies, contact tracing, and protection of the most vulnerable? Of course; assuming that we could get a high degree of buy-in (we can’t even get people to wear masks), that’s exactly what we are all hoping for. But that’s not ‘herd immunity’, and it’s clear that the cost in lives and suffering from a “just get it and get it over with” ‘strategy’ would be astronomical even with our most optimistic estimates. Trust me, I’m tired too; I completely understand the pull towards a roll the dice approach that just gets this over with and lets the chips fall where they may; that approach completely appeals to my intellectual and emotional fatigue. But the longer we can work together to flatten the curve, the more time we create to discover those new therapies, improve our understanding of the virus, and collect high quality data about transmission and vulnerability that can help us develop novel, strategic mitigation approaches (which would probably incorporate something like herd immunity); and we are already seeing the benefits of the work of this kind that we have done so far as a society.

Question #4: What is the RECOVERY Trial?

(Confession: nobody asked about this, but I’m going to write about it anyway)

The RECOVERY Trial is a randomized (poor British researchers spelled it wrong) clinical trial out of Oxford that has shown benefits from using low-dose dexamethasone (a cheap and readily available steroid) for hospitalized COVID-19 patients on oxygen or on a ventilator; you can read a more detailed analysis of the trial from First10EM. This is still in the peer review process but results have been incredibly promising; the study showed a relative decrease in mortality of 20% in hospitalized patients requiring oxygen, and up to a 35% decrease in patients requiring ventilator support. Unlike many of the drug therapies that have been touted up until now, this is based on a randomized trial and not on anecdotal evidence, so it is much more likely that these results will be reproducible when used broadly. Already this has become the standard of care in the hospitals in your city, and if we see these results persist with widespread use it has the potential to save tens or hundreds of thousands of lives. I wanted to write about it for two reasons. First, I want to call on us all now to not let this become the next hydroxychloroquine. The study has established the benefits of this drug therapy only in a specific group of people; hospitalized patients requiring oxygen or ventilator support. They also studied hospitalized patients who were not sick enough to need oxygen, and it showed no benefits whatsoever. There is no reason to infer that this medication is protective in those without severe symptoms or in asymptomatic individuals, and so there is no reason for individuals to ask their doctor for an outpatient prescription or for pharmacies or clinicians to stockpile the medication as we saw done with hydroxychloroquine. We can be thankful that we have at least one helpful medication for our sickest patients without that immediately translating into figuring out a way to get it for ourselves whether it would actually help us or not. And if peer review and follow-up studies and the increased clinical experience that comes with widespread use of dexamethasone ultimately shows that it actually isn’t helpful for COVID-19, that will be tragic; but we should all understand now that that is just how science works, and won’t be part of some big government conspiracy to prevent people from getting the medication, just as it wasn’t with hydroxychloroquine.

But even more importantly, I wanted to talk about the RECOVERY Trial because it illustrates exactly what it looks like to fight this virus by engaging in mitigation and flattening the curve. Since April people have been saying (and we have all been feeling, to some degree or another) that if a certain amount of death and suffering from the virus is inevitable, we might as well just get it over with. We have also heard the slightly more sophisticated position that as long as our hospitals aren’t overwhelmed and we aren’t running out of ventilators and other equipment and resources for sick patients, then we have reduced the danger as much as is helpful and anything more is unnecessary. The RECOVERY Trial is a powerful illustration of why flattening the curve is beneficial even beyond these important goals. If you had a severe case of COVID-19 one month ago and had to be on a ventilator, you would have been treated with hydroxychloroquine and not with dexamethasone; today, you would be treated with dex and not with hydroxychloroquine, and your chance of dying would be 35% less; and that doesn’t even take into account the less quantifiable benefits from all that your doctors have learned about this virus in the meantime. A month from now, with more high quality trials and more clinical experience, who knows what the new standard of care will be and how much better a very sick person’s odds of surviving the virus will be because of it. The reason I wear my PPE with every patient and am a stickler about fomites and transmission, the reason I wear my mask when I’m in public, and the reason I am writing from home instead of a coffee shop today and attended church online this morning, isn’t because I’m afraid of the virus; it’s because when and if (and for me it has always felt more like an ‘if’ than a ‘when’) I get COVID-19, I would rather be treated by doctors and nurses and respiratory therapists who have had ample time to learn how to fight it, who have perfected their approach to ventilator settings and other supportive techniques for this virus specifically, and who have access to medications that have been carefully studied and have been proven to be effective; and because I would like to have that knowledge base and those techniques and medications available if and when I have to treat you.

COVID-19 Questions and (attempts at) Answers, Part 2: What about antibody testing and asymptomatic cases?

June 29, 2020July 18, 2020 TJ WebbLeave a comment

Due to length, I’ve broken this post up into multiple parts.

Warning: These got really ‘mathy’ on me before I realized it was happening.

Part 1: Is the rise in cases just due to more testing?

**Part 3: Isn’t a surge a good thing? Herd Immunity and the RECOVERY Trial.**

Question #2: Do antibody testing and asymptomatic cases prove the virus isn’t as dangerous as we thought?

Asymptomatic Cases

The short answer here is, yes. And also in a very real sense… No. When antibody testing first began to confirm that a certain percentage of people contracted the virus but never developed symptoms, or had symptoms that were so mild they failed to associate them with the virus (‘weird how my allergies just acted for a couple of days’), it was great news for everyone. What it was not (and I’ve been on this soapbox for a while now) was proof that the ‘experts were wrong’ about how dangerous the virus is. I’ve been reading every model and study and expert opinion about COVID-19 I could keep up with for the past 3 months, and I cannot tell you the number of times that physicians and epidemiologists and researchers have either implied or explicitly stated that the mortality rates we were seeing from the virus didn’t account for asymptomatic and minimally symptomatic cases. I’m no expert, but I’ve typed it more times than I can count myself.

Actually I counted; it’s been 6 times. That’s still a lot.

Those scientists anticipated that a certain percentage of the population would contract the virus but never develop significant symptoms, but had to work from the best numbers they had until such testing was actually available. And it’s a very good thing that those assumptions were correct, since the original case fatality rates we were seeing were in the civilization ending range of 8-15% in certain countries. If antibody testing had been developed and found only a negligible amount of asymptomatic and minimally symptomatic cases, it would be devastating news for everyone; not least for the doctors, nurses, epidemiologists, and others who have turned their lives upside down to fight the pandemic. Accounting for asymptomatic and minimally symptomatic cases would clearly yield a much lower death rate, but still firmly in the very, very dangerous range. For instance, large scale antibody testing in New York in April found antibodies in 13.9% of the population (WBUR has an excellent article picking through the wildly varied estimates of asymptomatic cases) , which reduced their overall estimated fatality rate from 6% to 0.5%. Many current estimates place the overall fatality rate between 0.5% and 1.3%. For a virus this contagious, these are still scary numbers. Even here at the end of June, many people are still wanting to compare this to the flu to dismiss the danger, even though these much lower death rate estimates are still 5 to 13 times higher than seasonal influenza’s commonly accepted 0.1% fatality rate, and even though the flu itself regularly threatens to overwhelm our healthcare systems. Please keep in mind that this is at best an apples and oranges comparison. We don’t routinely measure influenza antibodies to determine the percentage of asymptomatic cases, focusing instead on those who are symptomatic, and our death rates for flu are based on a totally separate set of calculations (I talked about this in more detail in my response to the Bakersfield Urgent Care doctors). If you want to compare oranges to oranges we can look at excess mortality for both viruses. Consider the graph below from New York State: the first cluster of red crosses is the peak of the 2017-2018 flu season, the worst flu season I have experienced since starting medical school; the second is COVID-19 during New York’s surge in April.

Before we move on from asymptomatic cases, we need to mention two more things. First, while knowing the overall infection fatality rate including data from those who never had significant symptoms is great from an epidemiology standpoint, it doesn’t mean that the case fatality rate for people with symptoms is a ‘fake number’ or falsely elevated. If you develop symptoms and test positive for the virus, and especially if you end up in the hospital, it would be small comfort to know that some people didn’t get sick from it at all. We still need to know what the specific risk is for people with symptoms, and for people with severe symptoms, in order to properly counsel those patients and to inform our medical response. Second, asymptomatic cases are a double edged sword; yes, it means that some people will become immune without actually getting sick themselves, but it also means that some people can spread the virus without ever knowing they’ve had it. We all need to exercise caution even if we don’t have a cough and fever.

I realize this is the same joke from earlier. I just really like it.

Antibody Testing

One of the problems in determining a final overall death rate (besides the fact that we are still in the middle of the pandemic) is the accuracy of antibody testing, since we have to rely on this to tell us how many people had the virus and were either asymptomatic or didn’t get tested for it at the time. And this in turn relies on something called the positive predictive value, how likely it is your ‘positive’ test result has really detected the antibodies, which depends both on how well the antibody tests are designed (and their not being fake, which is apparently a proble m now as well), but also on the prevalence, or in this case cumulative incidence, of the virus. The higher the percentage of people who have actually had the virus, the more likely it is that a positive test really represents a true positive and not a laboratory error. It’s a relatively simple concept, but honestly it’s just unintuitive enough that I’ve struggled with it myself for years. Basically, every lab test has some degree of error; sometimes these tests will tell you that you have the antibodies when you don’t, and sometimes it will tell you that you don’t have them when you really do. The more rare the virus has been in your area, the more likely that your ‘positive’ test was the result of such an error instead of actually having the antibodies. Carry this to the logical conclusion; if you brought an antibody testing system back in time to last Summer when nobody had SARS-CoV-2 antibodies, or for that matter back to Medieval England, you would still have some tests turn positive; but they would clearly all be from laboratory error because the prevalence of the disease then would have been 0%. When doing these tests, we cannot ignore the importance of how common or rare the virus has been in the region where we are testing.

Still less useful than bringing Sony Walkman

Calculating positive predictive value based on prevalence can be done with just a few numbers (test sensitivity, test specificity, and prevalence) and the simple equation PPV = (sensitivity x prevalence) / [ (sensitivity x prevalence) + ((1 – specificity) x (1 – prevalence)) ] (Um, there’s also an online calculator if you’d rather follow along that way), and it’s always shocking to me how quickly the lab error for even very good tests becomes relevant when the prevalence of a disease is low. Most manufacturers rate their antibody tests in the extremely accurate range of 95-100% for both sensitivity and specificity (because of course they do); some have performed well in independent testing, but others not so much. Let’s use the online calculator (or the equation above, if you just really like that sort of thing) and plug in a few of these numbers.

Scenario 1: Post-Surge New York City, excellent quality antibody test.
- Let’s say you never definitively got diagnosed with COVID-19 during the surge in New York, and wanted to get an antibody test to see if you have already had it and are immune.
  - Sensitivity: 95% (.95)
  - Specificity: 95% (.95)
  - “Prevalence”: 20% (.2)
- Results: Positive Predictive Value = 82.6%
  - This means if you get a positive results from this very accurate test done after your city has survived a severe surge, there’s still about a 17% chance you don’t actually have the antibodies after all.

I hope you guys are having as much fun with this as I am.

Scenario 2: Pre-Surge Texas, excellent quality antibody test.
- Now let’s say you had the antibody test done a few weeks ago here in Texas, again with a test that has excellent accuracy.
  - Sensitivity: 95% (.95)
  - Specificity: 95% (.95)
  - “Prevalence”: 4.6% (0.046)
- Results: Positive Predictive Value = 47.8%
  - With a lower prevalence, a positive antibody test on the same machine is now about the same as a coin toss.

Scenario 3: Pre-Surge Texas, sub-par antibody test.
- Same scenario as the last, but the quality of the test isn’t quite as good as the manufacturer funded studies seemed to promise.
  - Sensitivity: 88.6% (.886)
  - Specificity: 90.2% (.902)
  - “Prevalence”: 4.6% (0.046)
- Results: Positive Predictive Value = 30.4%
  - At this point you are probably better off just switching the ‘positive’ and ‘negative’ labels on the readout…

Now, savvy statisticians will note three things in looking at the above numbers and playing around with the data. The first is that I’ve used the very antibody testing methods I’m questioning to fill in the prevalence, which is itself part of my calculations. Figuring out the real prevalence is a complex problem epidemiologists are still trying to solve; this is a simplification for illustrative purposes. But more importantly, you will notice that as the prevalence goes down so does the likelihood that a positive test was really positive; in fact, it drops quite precipitously, especially as you get below 5%. However, as the specificity– the likelihood that the test correctly calls a negative result negative– approaches 100%, the number of false positives actually drops to 0. If we want to make sure we never tell someone they are immune when they aren’t, we need a very high specificity; but because no test is truly perfect, this will mean some sacrifices in actually being able to detect the antibodies when they are there, which hurts our ability to accurately estimate the number of asymptomatic cases. To get a perfect specificity, you will lose some sensitivity, and vice versa; the right balance depends on what you intend to use the test for.

So all of that to say, when that antibody test you got comes back positive and the manufacturer says their test is “95% accurate,” you may be tricked into thinking it means there’s a 95% chance you really have already had the virus and now have antibodies against it. But they are only telling you half the story, and you either need access to some more data to make your calculations and determine the real positive predictive value, or at the very least you need to take it with a grain of salt and still exercise caution; especially if your area hasn’t actually had anything like a true surge yet. After all, only a great fool would accept what he was given, and you are not a great fool.

*Sorry, I’m not going to say “inconceivable.”*

COVID-19 Questions and (attempts at) Answers, Part 1: Is the rise in cases just due to more testing?

June 29, 2020July 18, 2020 TJ WebbLeave a comment

Due to length, I’ve broken this post up into multiple parts.

Warning: These got really ‘mathy’ on me before I realized it was happening.

Part 2: Do antibody testing and asymptomatic cases prove the virus isn’t as dangerous as we thought?

**Part 3: Isn’t a surge a good thing? Herd Immunity and the RECOVERY Trial.**

Question #1: Isn’t the rise in cases just a reflection of more widespread testing?

This is a question that has been on everyone’s minds since the very earliest days of our testing woes, back in March when we had barely any testing available. It has ranged from a very fair question to a rhetorical device for spreading misinformation, with at least one prominent political figure even seeming to say that it would be better if we didn’t test so much so that our numbers looked better. I honestly believe most people really are curious about the relationship between our testing numbers and our numbers of cases and are not asking to try to minimize the appearance of the surge we are facing in Texas right now. In one sense, we will always find more cases of a disease when we test for it than when we don’t; that’s a truism. But if we want to determine whether cases are really going up we can look at a few other parameters than the absolute number of positive tests that will inform our understanding of the ’75 new cases’ or ‘5,747 new cases’ we are seeing in the news and on social media each day (To go through these numbers I highly recommend you spend some time navigating the Texas DSHS COVID-19 Dashboard; both their case data and testing and hospital data sections).

The first number is the percentage of positive tests. Ever since testing became more widely available in April and we were able to shift away from testing only those with a high likelihood of having the disease and/or of developing complications, we have been testing essentially the same types of cases; people with some combination of cough, shortness of breath, fever, loss of taste and smell, etc. and/or known or suspected exposure to the virus. There are many causes of these types of symptoms, from allergies to other respiratory viruses to chronic conditions like asthma and COPD, and in our pre-surge days these explained the symptoms in the vast majority of people we tested. If you look at the Texas testing data from April you will see two things; an overall low number of tests (a very modest 5-10k per day) and a fairly high percentage of tests that are positive, between 10-14%. This reflects our very strict testing criteria at that time; we were only testing the people we already really thought had it. In late April and all through May we see an ever increasing number of daily tests and a falling rate of positive tests, a reflection of liberalizing testing criteria and strong evidence of overall low prevalence in our State. Throughout June, and especially over the last 2 weeks, we continue to see an increasing number of tests each day; but we are now also seeing our percentage of positive cases rising again. This isn’t because we’ve tightened up or restricted our testing criteria again; it’s because more people actually do have the virus.

This exactly matches my own clinical experiences; back in May I was testing for COVID-19 based on essentially the same criteria and clinical judgement I am using right now, but it was rare to get a positive case; I would know, because being told you have COVID-19 can be a very stressful experience, so I still personally call every patient I’ve tested who has a positive result in order to answer their questions and help them process that information. This past week I have had to make multiple of those phone calls daily and have been feeling the strain on my time that it has created. As a physician I was on the front lines in May just like I am now, and I can tell you that we are definitely feeling this surge in a way we didn’t then; it isn’t a statistical artifact.

The second kind of data that should inform our understanding of that increase in cases is the number of people who are hospitalized with COVID-19; and the number of people who are dying from it. A raw increase in cases without a change in the test positivity rate could certainly be explained by more widespread testing; but it could not explain why more people had severe enough symptoms to be hospitalized, and there is no question that we have seen an increase in hospitalized cases.

Many people will quickly point out that we don’t know what percentage of those people were hospitalized for COVID-19 related symptoms and what percentage just happened to have a positive test when they came to the hospital for other reasons. This is a seemingly fair argument on the surface, but it is guilty of two fallacies. First is the idea of COVID-19 infection being a coincidence that doesn’t effect the trajectory of someone’s chronic illnesses. For months now I have heard the argument that the people whom we know have the absolute highest risk of COVID-19 complications, elderly people with chronic heart and lung disease, have not died from COVID-19, just with COVID-19. Yes, they happened to have the virus but actually died, in large numbers, from their chronic illnesses all getting worse at the same time, during a surge in COVID-19 cases in their area. This is the tired conspiracy theory that doctors are misattributing the cause of death to inflate COVID-19 death numbers, and it’s one I’ve had to debunk over and over again on this blog; it willfully ignores the pathophysiology of the virus, the normal course of those illnesses, and the way that doctors understand and report contributing causes of death. The idea that we are suddenly seeing a huge uptick in COVID-19 hospitalizations as an artifact of testing patients when they come in and unrelated to the virus itself is just another version of that same conspiracy theory. It’s also a very hypocritical argument, considering the types of sources it is coming from. One of the criticisms about mitigation efforts from the beginning was that people who needed care might not come in to the clinic or hospital because of fear of the virus; it’s a very real concern and a problem I’ve fought against daily as a physician, and have been writing about since my earliest social media and blog posts during the pandemic.

Their argument has been that telling people that the virus is dangerous and taking mitigation measures would discourage them from seeking care for conditions that were really dangerous, like congestive heart failure or blood clots in the lungs, because they were more afraid of catching COVID-19 at the hospital. Our argument has been that the virus is dangerous, and that it also makes congestive heart failure more dangerous and actually causes blood clots in the lungs, so we have an obligation to keep people safe from the virus and help them navigate when and how to seek care for other health concerns; it’s work we are doing constantly in our clinics and hospitals. Now these same sources are arguing that in the midst of a dramatic increase in cases and our first real surge in Texas, thousands of people with conditions that put them at risk for complications from COVID-19 suddenly aren’t worried about the virus after all and are all seeking hospital care at the same time, and just happen to test positive for the virus while they are there. There may well be some situations where this actually is the case, and people who were overlooked by our healthcare systems really are now getting very sick from their diabetes or coronary artery disease at the same time as our surge (you can only ignore a worsening chronic illness for so long before hitting a crisis point), but the idea that this would happen on a broad scale, all at the same time, and that enough of those patients would be positive for COVID-19 that it would cause a state-wide spike in hospitalized virus cases is a very, very, frustratingly silly argument.

The final number we need to consider is the number of deaths, and here at least there is some good news; we are not seeing a substantial increase in the people in Texas dying from COVID-19, at least not yet. There are two ways to understand this. The optimistic way is to think that something has changed; either the virus has somehow become less deadly than before, or our increased understanding of COVID-19 has led to a better ability to fight the virus; improved disease-specific ventilation strategies, effective drug therapies, and more efficacious supportive care measures. In fact, there is a great deal of evidence that the latter really is true, as we will discuss in another post. But the pessimistic view (and the truth is probably a combination of both) is to realize that most people do not just get admitted to the hospital with severe COVID-19 infection and pass away the same day. There is a significant lag time as those patients are treated and fight against the virus, and our surge in hospitalized cases is only a little more than a week old.

Many of those hospitalized patients are fighting for their lives in the ICU right now, as the hospitals are starting to fill up around them and their nurses and doctors are becoming fatigued. Many of those people will recover, but many will not; and it will take a couple of weeks, and often times much longer, to see how many, and who. As we’ve seen elsewhere, the ratio of those who don’t recover will only increase if resources and the margin for careful attention and heroic efforts on their behalf begin to run short. Yes, our improved understanding of the virus and more effective therapies gives us a better chance to fight the virus than Italy had in March or New York had in April; but doubling down on the difficult work of mitigation now to prevent our healthcare systems from being overwhelmed in a couple of weeks when more and more patients reach their crisis point is every bit as important.