Expanding testing is very important now. It's also clearly an area we were unprepared. We should commit to having testing capacity long term, to both provide more certainty to anyone expanding testing capacity today, and to be prepared in the future.
I had some thoughts about how this testing would be best structured. It's not possible to test for an contagious disease you're not aware of, but much of the infrastructure for doing so can already be in place, ready to be adapted. That infrastructure would roughly boil down to
a) sample collection
b) sample handling
c) sample preparation
d) sample analysis
e) materials: reagents, etc.
Scaling these up from scratch is quite a bit more work than adapting to a new contagion. A commitment to having that infrastructure would have helped a lot with the current crisis.
Right now, the focus is rightfully on health care workers, suspected cases and essential workers. In terms of preparation though, in the early stages of an outbreak, the focal point of testing should start with International travel. As such, that's a good place to plan for future infrastructure to focus on.
Right now, the focus is on detecting COVID-19. During non-crisis times (which hopefully will turn out to be most of the time, as it has been in the past), testing capacity would be better off testing for a a number of known contagion's. I'm thinking of influenza itself as one of such case.
The overall impact here is that in addition to being prepared, we would also help reduce some other risks. Influenza deaths in the US are estimated as 12,000-60,000 per year on average. If testing at airports delayed the spread of new strains, or left some places unaffected, year over year, those would be significant improvements in well being.
The way I imagine this working is, you'd have sufficient capacity at airports to test all passengers for two contagion. During normal times, you'd select 10% of the passengers and test them each for 10 known contagions.
When a case among a departing passenger was found, you'd switch to testing all departing passengers for that. For those that test positive, you could tell them not to travel or take other precautions. How you would treat those found with positive tests would depend on a lot of factors, so I won't conjecture too greatly there.
When an arriving case was found, you'd start testing 100% of the passengers from that same departing location.. The departing location would hopefully be able to respond by testing all of their outgoing passengers. In addition, you'd want to test any returning travelers, as well as contacting and testing arrivals from that destination for the two week (or whatever period is appropriate for the identified contagion).
If this was a capability we had before this year, it's not outlandish to think we would have outbreaks in fewer places, that those would be smaller, and that the response would have been better directed and better prepared. The stable state value of the system would defend it against dismantlement, which is the fate of many preparedness stockpiles.
This screening network would have some value in normal times of reducing common contagions, but it's primary purpose would be to be ready to adapt to anything more serious. You'd have some idea of it's effectiveness by how well it reduced the spread of common contagions, like influenza? Could we wipe out Infuelenza in this way? Probably not, but it might save many lives despite this.
These thoughts are guided by two disciplines, economics and disaster planning for software infrastructure. The economics side tells me that if you want capital investment, you shouldn't make short term plans, you should think about long term incentives. It's a common mistake to think only about operating costs, i.e. how much does it cost to run a testing machine? But the cost of the machine is a big cost. Bigger than that is the cost of training for sample collection and preparation.
On the other side, my experiences with disaster planning for software infrastructure is that the most successful are actively used. Infrastructure that is held in reserve has a tendency to be less functional than advertised. Unless the costs of using it are very high, actively utilizing infrastructure validates that it's operational. Sure, you could run synthetic tests, but these are easy to overlook, whereas operational use is not so easy. You can (and should) still have excess capacity, but you do it by cycling it in an out of active use, by running it less than fully utilized, or by having lower priority work that can be terminated.
I offer the example of testing at airports, because they would provide the best early signal. Cities are connected to more total places by air travel than any other method. In 2019, the US had 925.5 million air passengers. Obviously going to be fewer this year, but lets assume it's in the same ballpark next year. That would be about 2.5 million passengers per day. If you take a sample from 10%, and test for 20 different contagions per sample, that's about 5 million tests per day, about what is currently recommended as a goal.
Depending on costs, how much capacity and how effective this system looks to be, there are other places you might look next, and if so, maybe you spread that capacity around more, taking a sample from 5% of air travelers normally.
I'm sure there are many ways this basic concept could be tweaked and improved upon, but the basic concept seems fairly sound. Objections might privacy concerns over being tested, or the cost... but honestly the cost-benefit ratio both in dollars spend and privacy is far better than that of our existing airport security.
I had some thoughts about how this testing would be best structured. It's not possible to test for an contagious disease you're not aware of, but much of the infrastructure for doing so can already be in place, ready to be adapted. That infrastructure would roughly boil down to
a) sample collection
b) sample handling
c) sample preparation
d) sample analysis
e) materials: reagents, etc.
Scaling these up from scratch is quite a bit more work than adapting to a new contagion. A commitment to having that infrastructure would have helped a lot with the current crisis.
Right now, the focus is rightfully on health care workers, suspected cases and essential workers. In terms of preparation though, in the early stages of an outbreak, the focal point of testing should start with International travel. As such, that's a good place to plan for future infrastructure to focus on.
Right now, the focus is on detecting COVID-19. During non-crisis times (which hopefully will turn out to be most of the time, as it has been in the past), testing capacity would be better off testing for a a number of known contagion's. I'm thinking of influenza itself as one of such case.
The overall impact here is that in addition to being prepared, we would also help reduce some other risks. Influenza deaths in the US are estimated as 12,000-60,000 per year on average. If testing at airports delayed the spread of new strains, or left some places unaffected, year over year, those would be significant improvements in well being.
The way I imagine this working is, you'd have sufficient capacity at airports to test all passengers for two contagion. During normal times, you'd select 10% of the passengers and test them each for 10 known contagions.
When a case among a departing passenger was found, you'd switch to testing all departing passengers for that. For those that test positive, you could tell them not to travel or take other precautions. How you would treat those found with positive tests would depend on a lot of factors, so I won't conjecture too greatly there.
When an arriving case was found, you'd start testing 100% of the passengers from that same departing location.. The departing location would hopefully be able to respond by testing all of their outgoing passengers. In addition, you'd want to test any returning travelers, as well as contacting and testing arrivals from that destination for the two week (or whatever period is appropriate for the identified contagion).
If this was a capability we had before this year, it's not outlandish to think we would have outbreaks in fewer places, that those would be smaller, and that the response would have been better directed and better prepared. The stable state value of the system would defend it against dismantlement, which is the fate of many preparedness stockpiles.
This screening network would have some value in normal times of reducing common contagions, but it's primary purpose would be to be ready to adapt to anything more serious. You'd have some idea of it's effectiveness by how well it reduced the spread of common contagions, like influenza? Could we wipe out Infuelenza in this way? Probably not, but it might save many lives despite this.
These thoughts are guided by two disciplines, economics and disaster planning for software infrastructure. The economics side tells me that if you want capital investment, you shouldn't make short term plans, you should think about long term incentives. It's a common mistake to think only about operating costs, i.e. how much does it cost to run a testing machine? But the cost of the machine is a big cost. Bigger than that is the cost of training for sample collection and preparation.
On the other side, my experiences with disaster planning for software infrastructure is that the most successful are actively used. Infrastructure that is held in reserve has a tendency to be less functional than advertised. Unless the costs of using it are very high, actively utilizing infrastructure validates that it's operational. Sure, you could run synthetic tests, but these are easy to overlook, whereas operational use is not so easy. You can (and should) still have excess capacity, but you do it by cycling it in an out of active use, by running it less than fully utilized, or by having lower priority work that can be terminated.
I offer the example of testing at airports, because they would provide the best early signal. Cities are connected to more total places by air travel than any other method. In 2019, the US had 925.5 million air passengers. Obviously going to be fewer this year, but lets assume it's in the same ballpark next year. That would be about 2.5 million passengers per day. If you take a sample from 10%, and test for 20 different contagions per sample, that's about 5 million tests per day, about what is currently recommended as a goal.
Depending on costs, how much capacity and how effective this system looks to be, there are other places you might look next, and if so, maybe you spread that capacity around more, taking a sample from 5% of air travelers normally.
I'm sure there are many ways this basic concept could be tweaked and improved upon, but the basic concept seems fairly sound. Objections might privacy concerns over being tested, or the cost... but honestly the cost-benefit ratio both in dollars spend and privacy is far better than that of our existing airport security.
Comments
Post a Comment