In the wake of the COVID-19 pandemic, an underutilized strategy for mass medical testing will save time, tests, and costs associated with reopening the United States. For more information email: email@example.com.
This article is by Jessica Adkins, M.D., an incoming resident physician in emergency medicine at the University of Kentucky; Matt Dawson, M.D., and Mike Mallin, M.D., both emergency medicine physicians and founders of Wild Health, a genomics based personalized medicine practice in Lexington; and
Taylor Bright, Ph.D., Director of Infectious Disease and Genomics Specialist at Wild Health.
By Jessica Adkins, M.D., Mike Mallin M.D., Taylor Bright, Ph.D.,
Matt Dawson, M.D. – firstname.lastname@example.org
Since the rise of COVID-19 in the United States, test shortages have been the Achilles heel of pandemic response. Even after months of advancements in distribution of SARS-CoV-2 tests and supplies, testing continues to lag behind the growing need. Demand has exploded, not only because new cases continue to rise and must be diagnosed, but also because RT-PCR viral testing for healthy people is a keystone in plans to reopen society. To meet this need, some scientists and physicians advocate dramatically reducing the time and cost of screening using a strategy known as pooled testing.
Pooled testing works best in screening large populations that have relatively few infections, as is the case in most of the United States. With this method, a portion of the material in people’s samples are “pooled” into groups and run as a single PCR test. When negative, all people whose samples are in the pool are cleared. When positive, leftovers from their initial samples are retested one by one.
Pooled testing was born during World War II, when the U.S. Public Health Service and the Selective Service System were tasked with screening 10 million military inductees for syphilis. At that time, syphilis had a low prevalence, meaning it was a relatively rare disease among these men. In response to this challenge, mathematician Robert Dorfman pioneered the elegant strategy now known as pooled testing. He then optimized it by defining equations for the ideal pool size for any disease prevalence and sample size.
For an example of Dorfman’s strategy in action, consider a population with a known coronavirus infection prevalence of 1% (meaning 1% are infected at the time of the study). If 300 people there are tested, likely 3 would be infected. Using individual testing, one would need to perform 300 tests on an RT-PCR machine. Instead, if one round of pooled testing is implemented, it is possible to test them in just 60 tests.
First, the 300 people are partitioned into 30 testing pools of 10 people per pool. When the pools are tested, the 3 infected people would make 3 out of the 30 pools positive. The other 27 testing pools are negative, so 270 people are cleared by only running 30 tests. For the second round of testing, the 30 people in the 3 pools that tested positive are tested individually to find the 3 infected people. This screens the same number of people using 20% of the tests.
In some cases these savings can be improved with multiple rounds of dividing into pools, testing, and again dividing the positive pools into subgroups. For example, imagine screening 9000 people who have a 0.5% prevalence of disease, meaning there are only 45 infected people in the large group. Instead of individually testing them with 9000 tests, pooled testing would use 1275 tests. If instead, after the positive pools were found, they were subdivided into even smaller pools for retesting, total tests could be further reduced to 960 tests. This requires about 11% of the 9000 tests needed for individual testing. For a detailed explanation of these examples, see the Wild Health white paper Pooled Testing.
In the United States, the true prevalence of coronavirus is uncertain. However, it seems to be low enough for pooled testing to work. In the largest seroprevalence study of COVID-19 to date, Stanford researchers tested 5,754 Major League Baseball employees for an immune response indicating past exposure to the coronavirus. Only 0.7% had positive serology test results. It is reasonable to assume that if RT-PCR swab testing had been done, an even smaller minority of people would show current infection. Even though this sample of MLB employees does not represent the entire U.S. population, it is reassuring that prevalence of active infection is likely to be under 5%, where pooled testing provides the most benefit.
Savings that are possible when screening different sizes of low-prevalence groups using pooled testing instead of individual testing. There are significant savings, reducing total tests by more than 70%, when less than 2% of the population is infected.
Tests can be saved by using pooled testing to screen populations with prevalence of disease between 0.1% and 25%. Pooled testing reduces the number of tests by more than half when the group has a prevalence under 5%, as in asymptomatic people in the U.S. today. For more information, see the Wild Health guide Pooled Testing.
This relationship between lower prevalence and better savings partially explains why pooled testing has been so slow to take hold during the COVID-19 pandemic. Until the past few weeks, testing was in such short supply that it was limited to patients with symptoms who were seeking emergency treatment. In some areas, coronavirus prevalence among those patients has been over 10%. Though pooled testing could conserve tests for populations with a prevalence of 10-25%, the margin of savings is narrower, and individual testing was a simpler option for overburdened public health agencies.
Now, healthcare systems are going beyond testing the sick. They have begun to screen representative samples of the community, including those with and without symptoms. The purpose is to estimate the number of coronavirus infections in the whole population. Over time, this paints a picture of the virus’s movement through the community. Is it improving? Is it worsening? Where can we focus our resources to better protect our citizens? Pooled testing has already allowed public health agencies in Nebraska and India to test these higher-level questions more efficiently.
Screening samples of the U.S. population regardless of symptoms is part of the federal government’s master plan for reopening, with the reported goal of 2% of the population per state per month. This equates to 218,000 tests nationwide per day. When averaged across the country, the U.S. is finally meeting this goal. However, leading ethicists and economists say this is not truly enough to properly monitor the pandemic. Furthermore, states vary widely in the testing rates they have achieved, partly due to the high cost of individual testing.
On a smaller scale, demand is also surging in the reopening of workplaces and small events. By testing employees before their return to work, businesses can drastically reduce the chance of a workplace outbreak. Unfortunately, businesses small and large struggle to provide viral RT-PCR testing at the price of $100 per employee, which is the average market price. The price and volume of individual tests are major barriers in reopening the United States.
When testing is more efficient and widespread, the use of tests on workplaces and sporting events is much easier to justify. Upon the return of professional sports, for example, athletes that cannot practice social distancing during play will need frequent testing. Before a game, the most important question is simply: “Are any of these people infected?” This could be checked by swabbing the noses of athletes and essential staff the day before a game. Half of each sample would be transferred to a single pool for the whole team. If the pool is positive, the remaining sample material would be retested individually to identify the infected person.
Pooled testing could even allow for improved accuracy in high-risk groups. If fewer tests are required to test a population, some groups could be tested twice to reduce false negative results, or more frequently to identify infections earlier and reduce the spread of disease. This could be targeted toward outbreak-prone settings like nursing homes, correctional facilities, and sports teams.
As the United States recovers from the COVID-19 pandemic, healthy people’s access to mass testing may serve as their key to workplaces, schools, and, eventually, sports. More than ever, we must be strategic about our use of laboratory resources. This 80-year-old testing strategy was developed to protect the Greatest Generation from infectious disease as they fought for freedom and defended the vulnerable amid World War II. Today, it better equips us to defend them and their children, offering peace of mind as we rebuild our economy and way of life.
1. Jordan Ellenberg. “Five People. One Test. This Is How You Get There.” The New York Times, 7 May 2020, https://www.nytimes.com/2020/05/07/opinion/coronavirus-group-testing.html
2. Centers for Disease Control and Prevention. “COVIDView: A Weekly Surveillance Summary of U.S. COVID-19 Activity.” Centers for Disease Control and Prevention, 8 May 2020, www.cdc.gov/coronavirus/2019-ncov/covid-data/covidview/index.html.
3. Stein, Rob, et al. “U.S. Coronavirus Testing Still Falls Short. How’s Your State Doing?” NPR, NPR, 7 May 2020, https://www.npr.org/sections/health-shots/2020/05/07/851610771/u-s-coronavirus-testing-still-falls-short-hows-your-state-doing.
4. Dorfman, Robert. The detection of defective members of large populations. The Annals of Mathematical Statistics, 14(4):436–440, 1943. doi: 10.1214/aoms/1177731363.
5. Duff-Brown, Beth. “Study Finds .07% COVID-19 Antibodies Among MLB Employees.” Stanford University News, Stanford University, 12 May 2020, fsi.stanford.edu/news/study-finds-07-covid-19-antibodies-among-mlb-employees.
6. Vergun, David. “First Peacetime Draft Enacted Just Before World War II.” U.S. Department of Defense, 7 Apr. 2020, www.defense.gov/Explore/Features/Story/Article/2140942/first-peacetime-draft-enacted-just-before-world-war-ii/.
7. Aldridge, Matthew, et al. “Group Testing: an Information Theory Perspective.” ArXiv.org, Foundations and Trends in Communications and Information Theory: Vol. 15: No. 3-4, Pp 196-392, 2019, 30 Sept. 2019, arxiv.org/abs/1902.06002.
8. Hogan, CA, Sahoo MK, Pinsky BA. Sample Pooling as a Strategy to Detect Community Transmission of SARS-CoV-2. JAMA. Published online April 06, 2020. doi:10.1001/jama.2020.5445
9. Lohse, Stefan, et al. “Pooling of Samples for Testing for SARS-CoV-2 in Asymptomatic People.” The Lancet Infectious Diseases, 28 Apr. 2020, doi:10.1016/s1473-3099(20)30362-5.