Read on Twitter
I'm seeing more and more suggestions that contact tracing and/or physical distancing isn't needed and we could solve COVID-19 with widespread testing alone. E.g. just test everyone once a week/fortnight to get R<1. Sounds straightforward? Unfortunately not... 1/
On average, each person infected with COVID-19 can spread infection to around 3 others in the absence of control measures. So if we want to get fully back to normal, testing will need to stop at least 2/3 of transmission (so each case infects less than one other person). 2/
Even if we ignore the (enormous) logistical challenges of pulling together resources and expertise to run the required millions of daily PCR tests, we still have to remove 2/3 of the opportunities for transmission. So let's suppose everyone receives a weekly test. What next? 3/
Testing for COVID-19 is generally done by nasopharyngeal swab, which is not easy to self-administer. But let's assume an easier nasal self-swab is possible at scale (e.g. https://www.medrxiv.org/content/10.1101/2020.04.09.20057901v1) and 95% of people attempt the test each week. 4/
Of course, people also have to collect the sample properly, pack it and ship to a lab to be tested. Let's suppose 95% manage to do this. (For context, this study found less than 80% of participants correctly collected and shipped a nasal swab): https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4653956/) 5/
What happens if someone is infected? Once the sample arrives at the lab, the PCR test then needs to detect the virus. At best, it might have around a 95% chance of doing this (i.e. sensitivity = 95%) https://www.medrxiv.org/content/10.1101/2020.04.05.20053355v2 6/
So to recap, our optimistic assumptions have 95% of people doing the test, 95% doing it properly, and 95% of infections being picked up. Multiplying together, that means we'll detect around 86% of infections. But there's another problem... 7/
We don't just need to detect infections, we need to reduce transmission. If testing is weekly, and people become infectious after 3-4 days, then (assuming same-day turnaround of results) we'd only detect half of cases before they became infectious https://www.nature.com/articles/s41591-020-0869-5 8/
Optimistically, let's assume successful detection prevents 75% of onward transmission on average. We've estimated we can detect 86% of infections, so multiplying together that means preventing just under 65% of transmission.... 9/
But remember, we need to stop at least 2/3 of transmission to control COVID (and even then it won't disappear immediately). Unfortunately, even our very optimistic scenario falls short of this - if mass testing were feasible, it would prob need to be biweekly, not weekly. 10/
Inefficient population testing could also crowd out better targeted approaches, such as contact tracing, which can informed by we know about transmission chains and - as we get more data - which contacts/environments are particularly risky 11/
We really do need better testing/tracing strategies and other innovations to reduce COVID-19 transmission. But this figure shows the fundamental challenge we face with COVID control - this is the problem any strategy has to solve: https://twitter.com/trvrb/status/1250855521327788032?s=20 12/12
