In my column on Wired, I expanded on something I first mentioned here a few days ago when I heard Google is building a "triage tool" to determine who will get tested for the novel coronavirus.
The key idea is that tests provide invaluable data to set public-health policy. Data is the antidote to uncertainty which is the biggest problem facing the world in the early stages of a pandemic. It is important to have both more data and better data. The timid approach to testing in the U.S. and the U.K. so far is producing a limited dataset plus a highly biased dataset.
See the column at Wired: The Problem with Trump's Triage Testing
A few new points I made in this piece: the explore-versus-exploit tradeoff, and the blind spot of asymptomatic carriers of the virus, the role of predictive models
***
Further reading:
The NY Times has a piece out on how far behind the U.S. is in terms of testing
New Scientist qutoes WHO calling for more testing
Vox thinks the U.S. lockdown will take a long time, maybe a year. Important quote: "How are you supposed to implement effective containment measures if you don’t know the actual prevalence in the population? And we still don’t." Because of insufficient and triaged testing!
[3/19/2020] Reporters are finding examples of places that do rigorous testing and track down the positive cases.
The town of Vo near Venice (link)
A contrast of two neighboring counties, Lodi and Bergamo, in Lombardy (link)
It is of course impossible to test 300 million people all at once. That's not the point. You find the clusters, and within those towns, test extensively, and isolate positive cases, whether or not they have symptoms. According to the case studies, that's how they "flattened the curve" which is really just saying containing the spread.
No country has tried the test everyone strategy except as part of research. The reason is that it is unnecessary and it is expensive. The way NSW, the state I live in Australia, is working now is to test anyone who is admitted to hospital with a respiratory condition. They are also testing anyone with a respiratory symptoms that are in an area with unidentified transmission. Then for identified cases the contact tracing starts and anyone who has been in contact with a case is also tested. Given that our active cases out in the community are probably in the order or less than 5 per 10,000, testing of the asymptomatic population or even people with respiratory symptoms in many areas is not effective use of tests. One thing we haven't been doing enough of is convincing people with mild respiratory symptoms to self isolate, and if possible their households as well. It is cheap, and given the number of false negatives, almost as effective as testing. The advantage of testing is that it justifies contact tracing. When we have the testing capacity we should be doing everyone with respiratory symptoms.
Posted by: Ken | 04/13/2020 at 12:36 AM
Between when I wrote that article and today, the case has only strengthened. Evidence is piling up that one of the reasons why this novel coronavirus is so potent is that it spreads fast, and one key reason for the speed is the prevalence of silent carriers - people who do not show any symptoms but have high viral load. One of the gravest mistakes made here is testing only people with severe symptoms. We can't even diagnose the problem of silent carriers, let alone solve it. It's tragic.
Posted by: Kaiser | 04/13/2020 at 02:07 AM