Using random testing to manage a safe exit from the COVID-19 lockdown

Markus Mueller; Peter Derlet; Christopher Mudry; Gabriel Aeppli

for the testing rate r. Note the inverse proportionality to the parameter i * , for which Eq. (15d) is a conservative estimate. Using this value yields an estimate of the order of magnitude required for Switzerland. In the next section we simulate a full mitigation strategy and confirm that with additional capacity for just about 15'000 random infection tests per day a nation-wide, safe reboot can be envisioned for Switzerland. We close with two observations. First, this minimal testing frequency is just twice the testing frequency presently available for suspected infections and medical staff in Switzerland. Second, while the latter tests require a high sensitivity with as few false negatives as possible, random testing can very well be carried out with tests of lower quality in that respect. Indeed, an increase in false negatives acts as a systematic error in the estimate of the infected fraction of people, which, however, drops out in the determination of its growth rate, 10 as long as the fraction i is not close to 1. However, the success of random testing does rely on a very low probability ( i * ) of false positives (as is the case of current PCR tests). Otherwise the signal from true positives would rapidly be overwhelmed by the noise from false positives.