Increase efficiency of testing for COVID-19

Background

The lack of reliable testing for COVID-19 has been a huge hurdle to the US’s response to the pandemic. [1] Washington State has taking drastic actions to slow the spread of the virus, including closing all schools, bars, gyms, and restaurants. According to The Atlantic, experts now believe that massive-scale testing will be necessary to restore normal economic conditions. [1]

Private companies such as Quest Diagnostics, LabCorp, and Roche are beginning to ramp up testing capabilities – for instance, LabCorp recently announced that it is able to run 20,000 tests per day [2]. However, testing remains a major bottleneck in the response for COVID-19. Grouping samples for testing can reduce the number of test kits needed to perform the same number of tests, allowing up to 5-10x the amount of people to be tested with the same number of test kits.

Recommendations

Group samples for testing by household. If one member of a household has COVID-19, it is extremely likely that other members of the household have it; testing multiple members of the household is redundant.

For samples from different households, the number of samples to group together can be optimized based on the prevalence of the virus in the population – see appendix A for further details.

1000 Test Kits
Probability of Positive Group Size Extra People Tested
A: 0.3% 20 8232
B: 0.4%-0.5% 15 6573
C: 0.6%-0.7% 13 5322
D: 0.8%-0.9% 12 4535
E: 1% - 1.2% 10 3886
F: 1.3% - 1.6% 9 3269
G: 1.7%-2.1% 8 2742
H: 2.2% -2.9% 7 2245
I: 3.0% - 4.1% 6 1766
J: 4.2% - 6.6% 5 1263
K: 6.7% - 12.4% 4 768
L: 12.5% -20% 3 334
M: 20.1% -25% 3 141
People 1000 Total # of Tests Needed Based on the Probability of Having the Virus
Group Size N-Group 0.30% 0.50% 1% 5% 6.50% 10% 20%
1 1000 1000 1000 1000 1000 1000 1000 1000
2 500 506 510 520 598 626 690 860
3 333 342 348 363 476 516 604 821
4 250 262 270 289 435 486 594 840
5 200 215 225 249 426 485 610 872
6 167 185 196 225 432 499 635 905
7 143 164 177 211 445 518 665 933
8 125 149 164 202 462 541 695 957
9 111 138 155 198 481 565 724 977
10 100 130 149 196 501 589 751 993
11 91 123 145 196 522 613 777 1005
12 83 119 142 197 543 637 801 1015
13 77 115 140 199 564 660 823 1022
14 71 113 139 203 584 681 843 1027
15 67 111 139 207 603 702 861 1031
16 63 109 140 211 622 721 877 1034
17 59 109 141 216 641 740 892 1036
18 56 108 142 221 658 757 905 1038
19 53 108 143 226 675 774 918 1038
20 50 108 145 232 692 789 928 1038
21 48 109 148 238 707 804 938 1038
22 45 109 150 244 722 817 947 1038
23 43 110 152 250 736 830 955 1038
24 42 111 155 256 750 842 962 1037
25 40 112 158 262 763 854 968 1036

Sources:

[1] https://www.theatlantic.com/health/archive/2020/03/how-many-americans-are-sick-lost-february/608521/

[2] https://www.labcorp.com/information-labcorp-about-coronavirus-disease-2019-covid-19

2 Likes