This past Sunday I woke up and did not feel well. I had a slight headache, some body aches, and was generally tired. I went about my morning routine with the kids, and within half an hour my headache was worse, and I developed cold chills and a mild dry cough. When I took a deep breath, I felt a bit of pressure in my chest.
My wife and I quickly arranged for me to be isolated in our basement, and I got in bed and took my temperature, which was 102° F. On paper, I had most of the COVID-19 symptoms, so I called my primary care doctor’s office, and was routed to MGH’s COVID-19 line. I went through their questionnaire and was told that in all likelihood I had COVID-19, but that due to limited testing capacity, I was not eligible to be tested.
I’ll come back to the lack of testing in a moment. But first, I feel very fortunate to report that whatever this was, it seems to have been mild. The rest of Sunday was not fun, and overnight was worse, but by Monday morning almost all of my symptoms were less severe and my fever was down to 100° F. By Monday afternoon my temperature was back to normal and has stayed that way.
I had a lot of questions for the nice woman on the MGH COVID-19 line about why testing wasn’t available for me. Not because I felt I needed a test – I’m low-risk and it wouldn’t have changed my plan to hydrate, rest, and stay away from my family – but because I had previously been following the outbreak statistics closely, and was under the impression that testing availability in Massachusetts had improved substantially. And testing is obviously extremely important for getting good data on what’s actually happening, putting contact tracing programs in place, and ultimately getting this whole thing under control.
I remembered in mid-March Governor Baker announced a goal of scaling daily tests from less than a thousand per day to 3,500. And then within a week, Massachusetts blew through that goal and was consistently doing 5-6k tests per day.
And yet I was told over the phone that MGH was generally only doing tests for:
- Individuals being admitted to the hospital
- Individuals with symptoms AND risk factors such as old age or other conditions known to increase the severity of COVID-19
- Medical professionals and first responders
I subsequently checked if I could make an online appointment for on of the drive-up testing sites, and received the same response.
In a limited testing capacity situation, this seems to be exactly the right thing to do. But it does mean that (1) actual cases are still being grossly under-reported, and (2) testing capacity is probably an order of magnitude lower than it needs to be.
I knew both of these were likely true before this experience, but it helped highlight the extent of the issue.
In mid-March everyone knew the data was significantly distorted by testing practices and the lack of capacity. But at this point I’d hoped that most people who were told by a medical professional that their symptoms matched COVID-19 would be able to get a test, and ideally we’d already be doing random testing to identify asymptomatic individuals as Iceland is doing.
Nate Silver had a nice piece on this – Coronavirus Case Counts are Meaningless. He uses four different testing scenarios in a made up country called Covidia to explain how testing capacity can both distort the understanding of what’s actually happening, as well as the right policy response. He highlights how small differences in policy responses and their timing can have significant differences in infection rates and deaths.
So what does it mean if we’re still only testing high-risk individuals? Here are a few thoughts, none are particularly original, but all are relevant to thinking about what has to happen next:
- Were significantly under-counting cases. If basically nobody under the age of 55 is tested unless they are high-risk or show up at a hospital, and young people are (1) much more likely to disregard social distancing recommendations, and (2) much less likely to need to go to the hospital when infected, I’d guess the real case count is 5-10x what we’re seeing. Possibly more if asymptomatic cases are as common as some reports are suggesting.
- We’re overstating severe case rates and death rates. If we don’t have a true handle on how many infections there are, the data will not be useful in getting an accurate estimate of the true rate of bad outcomes. Maybe that’s less important now when it’s more critical to estimate the absolute number of severe case hospital admissions, but it will be very important down the road when assessing risk and evaluating longer term policies.
- We don’t know where new infections are growing the fastest. This has important implications on social distancing policy, and where it needs to be stricter. Getting good at this seems critical to preventing a second wave once we get through this first one.
- We’re not yet equipped to properly conduct contact tracing. I understand Massachusetts has launched an initiative to begin tracing, which is great, but they’ll likely only be tracing the contacts of a small subset of infected individuals. But that’s not a reason to delay this, as first phases will likely have some impact and will provide valuable leanings that can inform improvements and help scale the program.
So what should happen next?
There’s been a lot of talk about the lack of competence in various levels of government. I agree with most of it. And yet it’s also clear that some state and local governments are doing an outstanding job, despite minimal support from the Federal government. I’d put Massachusetts in that group, all things considered.
At this moment in time, I’m not sure how useful it is to dwell on exactly how incompetent the incompetent people and agencies are. We can do that later. I’m fairly certain they won’t become competent over the next 8-12 weeks, which will be the most critical time to lay the groundwork for what society and the economy look like over the next year.
My take is that we still aren’t throwing enough money at the problem. At least in the right places.
Since Sunday, I’ve been encouraged to rest and work a bit less. Guidance that I’ve mostly followed. So yesterday I decided to review in more detail what’s in the $2.2 trillion stimulus bill. Here’s the simplest visualization I found:
Almost all of the money goes towards treating the symptoms of social distancing (e.g., keeping businesses afloat, unemployment benefits), which is important, but almost none towards eliminating the virus itself. We need to be doing both, otherwise social distancing will extend much longer, the economic costs will be much greater, and more people will likely die.
Why aren’t there hundreds of billions earmarked for scaling up testing so that every American can be tested multiple times, accelerating vaccine development (Bill Gates is obviously a hero in all this, but why shouldn’t the government be paying to build factories to manufacture all the vaccine candidates?), and deploying massive contact tracing programs leveraging a mix of technology (i.e., smartphones) and providing jobs to many of the newly unemployed people?
It seems that large amounts of money focused on eliminating the virus could be spent in some reasonably obvious ways:
Infection Testing. The technology to test if someone has COVID-19 exists. This seems to be a logistics/operations problem – how you scale and deploy that technology at an unprecedented pace? Smart people are working hard at this and doing incredible things – the US went from 25k tests per day three weeks ago to 135k today – but this clearly needs to be scaled more quickly. What would it cost to get to 20 million tests per day, and what creative solutions – potentially leveraging the Defense Production Act – could help get there yesterday? There’s been a lot of talk about using this law for the production of ventilators, but why not tests? We need to first be able to test anyone with any symptoms whatsoever, but then quickly pivot to randomly testing a portion of the population. It seems that an ideal solution – one that would facilitate some loosening of social distancing – would be to test everyone in the country that hasn’t already had COVID-19 every couple weeks until a vaccine is available.
Serological/Antibody Testing. Similar to infection testing, this technology exists and these tests are beginning to roll out in the US and many countries around the world. These tests will tell us who has already had the infection so that they can (1) confidently stop social distancing, and (2) potentially donate blood with antibodies to help those who have the infection. These tests also need to be scaled rapidly to a very large number.
Vaccines. Lots of people keep asking why vaccines take so long, and credible experts keep repeating that it’s likely going to be 12-18 months before a mainstream vaccine is available. I’ll be honest, I trust them, but I still don’t understand why. The progression seems to be: research, trials, mass manufacturing. Many vaccine candidates are already in the trial phase. What could be done to run elements of this in parallel, and what shortcuts could be taken that would be an acceptable risk given the severity of the virus? Also, it’s clear from Bill Gates’ announcement that you can expedite the whole process by starting to build manufacturing capacity for candidates before they are proven, with the implication being that the investment will be wasted if the candidate trial fails (or if it proves to be less compelling than another option). When the global economy is losing trillions of dollars, it’s smart to risk a few billion on this. But could even more funding build this capacity faster, or at a larger scale? Why not match the few billion Bill Gates is putting into this with another $10 billion? I read an article earlier about how in 1957 a US microbiologist was alarmed about a novel strain of the flu, and helped to develop a vaccine and manufacture 40 million doses in 4-5 months. That was over 60 years ago. I appreciate the technological challenges may be different, but surely there must be a way to match or beat that speed today when the government is in the middle of spending 10%+ of annual GDP on stimulus?
Contact Tracing. I’ve caught wind of some large technology firms being hired to design and implement these programs at the state level. To get this right, you probably do need lots of different experiments to nail down the approach, logistics, technology, and privacy. But the whole thing could move a lot faster if the Federal government awarded 10 or so pilot contracts at the state level, and promised the company with the most effective solution a contract to roll something out nationally.
Better Data: The states are all beginning to report data in a similar way, and there are many great academic, non-profit, and other web resources pulling together dashboards/tracking to help understand what is going on. But why not fund a much more comprehensive effort? This is basically the CDC’s job, but if the agency isn’t up to the task, outsource it. In addition to test results, hospitalizations, and deaths, there are many other things that should be tracked. Take me for example. I’m now in an MGH database, likely flagged as a clinical diagnosis of COVID-19. That’s not as accurate as a test, but it’s better than nothing. For every positive test there are probably many more people who called their doctor, reported COVID-19 symptoms, and were told to stay home unless they have trouble breathing. Why can’t a standard best practice for collecting and storing this information be defined, and then have the information be aggregated and reported for the entire US?
Congress is beginning to plan the next round of stimulus. I don’t mean to diminish the need for more support for individuals, families, and businesses. These should all be prioritized again. But I’d like to see some of the above items also prioritized in the same bill, such that meaningful progress can be made in the next 8-12 weeks and we can get this done with one more bill.