Research highlight

Infectious diseases: Reducing the cost of large-scale COVID-19 testing


October 21, 2020

A mathematical approach to screening large populations for SARS-CoV-2 infections using small numbers of tests is described in Nature. This method — which is more suitable for monitoring infections on a large scale, rather than for confirming cases in clinical settings — may be particularly beneficial in low-income countries that cannot afford large numbers of tests.

Rapidly identifying and isolating individuals infected with SARS-CoV-2 is an important strategy to control the spread of this virus. PCR tests, which can detect viral RNA, are very accurate but cost around US$30–50 per test. These costs are particularly challenging for low-income countries.

Combining (or pooling) samples and testing them together can increase the efficiency of the analysis and reduce costs. This is particularly effective when diseases have a low prevalence across populations and the majority of tests return a negative result. The standard group testing procedure works by combining samples into groups of a certain size before testing. If a group tests positive, the samples in the group are then tested one-by-one to identify positive individuals.

Wilfred Ndifon, Neil Turok and colleagues present a mathematical algorithm to optimize the screening of SARS-CoV-2 infections through using pooled, parallel PCR tests. The basis of the approach is to take a group of samples, split them into sub-samples and recombine them as if they were in a high-dimensional cube. If there is a single sample that is positive for SARS-CoV-2, the coordinates on the cube that correspond to positive pools will immediately identify it. Otherwise, additional rounds may be necessary, as sub-samples are re-arranged using the algorithm. The size of the sub-groups can be optimized; groups can be made larger with decreasing prevalence, to ensure that two rounds of testing will typically be required.

doi: 10.1038/s41586-020-2885-5

Return to research highlights

PrivacyMark System