Tags
While pooling samples might make COVID-19 testing more efficient (1), a one size fits all approach wouldn’t work. To generate useful information, each pooling effort first requires empirically working out critical kinks, some general and others unique. Test here refers to RT-PCR for detecting SARS-CoV-2 viral RNA.
Advantages of pooled or group testing are obvious
- Save time, resources and equipment.
- Do fewer tests and yet cover a larger proportion of the population.
- When a pooled sample is negative, all the individual samples in it are likewise deemed negative. No more need to test them separately. Thus need to run fewer tests to estimate infection prevalence within a given population at a given point in time.
Disadvantages of pooled or group testing
- According to statisticians, pooled or group testing isn’t more efficient than individual testing when prevalence is high, say >30% (1, 2). In practical terms, this means lower the prevalence, greater the value of pooled or group testing.
- Accounting for reduced test sensitivity could be problematic. Samples from individuals with low viral loads would get diluted even further when pooled with those who are virus negative, increasing the likelihood that the former could be erroneously reported as negative, i.e., false negatives.
- For example, if ‘true’ asymptomatics (as opposed to pre-symptomatics) harbored lower viral loads, happenstance pools of asymptomatics and true negatives would reduce the chance of accurately detecting the former.
I. General kinks
Optimal group size: Has to be worked out individually for each available RT-PCR test and diagnostic system designed to detect SARS-CoV-2 virus to ensure pooling/grouping doesn’t compromise test sensitivity.
Sample material choice: Pool what material? Clinical samples or the purified RNA extracted from them?
Most published or preprint SARS-CoV-2 papers have compared pooling of either clinical sample (material from naso or oropharyngeal swabs or their lysates) or the RNA extracted from them to the respective individual sample materials.
An Ethiopian preprint compared both types of sample materials side by side and found it could detect positive SARS-CoV-2 from clinical samples pooled 8:1 (8 negatives pooled with 1 positive) but could detect it at a lower dilution of 10:1 when looking at RNA pools (3), suggesting RNA extract pooling may retain greater test sensitivity, an unsurprising result since clinical samples often contain biological substances such as enzymes that are natural inhibitors of the PCR reaction. However, choosing between clinical samples or their lysates versus RNA extracted from them isn’t easy since both approaches have their advantages and disadvantages.
- Given human genetic diversity, some individuals would likely have higher baseline concentrations of such inhibitory substances in their mucosal secretions compared to others. Pooling in such cases could inadvertently expand their inhibitory potential leading to higher risk of false negatives especially when pooling among samples with low viral loads.
- On the other hand, pooling purified RNA extracts is more expensive because extracting and purifying RNA from each individual clinical sample is more resource intensive compared to doing the same from one pooled one.
- Bottom line, choosing which sample material to pool comes with different price tags and could have outsize consequences on test results. Budgets would play a role. So would the dynamics of the outbreak within a population being considered for pooled/group testing.
- For example, while clinical sample pooling might be a safe choice for testing older residents in a skilled nursing facility, assuming both greater prevalence as well as higher individual viral loads, purified RNA extract pooling might be a better choice for testing among younger college students, many of whom might be more likely to be asymptomatic with lower viral loads.
II. Specific kinks
Optimal group size: How many samples to pool? That would vary from place to place as well as from time to time within a given place, being unique to each place based on prevailing local virus prevalence at a given time.
Sample source choice: Pool which samples? Just a bunch collected from random people or should there be some method to the madness? A few different types of pooling or grouping find value in this regard,
- Cohort pooling: choosing to pool samples from people in a way that maximizes the ability to detect maximal number of infections using minimal number of tests, that should be the goal of any cohort pooling effort.
- A simple rule of thumb would be close physical proximity to one another for protracted periods of time on a regular, e.g., daily, basis. Those living in the same household or working in the same physical space are obvious candidates. It could also be those living in the same apartment block, same floor, etc. The inherent logic to the choice of cohort that lends meaning to pooling or grouping choice in each and every instance should be maximizing infection detection with minimal tests.
- Sub-pooling: Consider a hypothetical. Let’s say 15 potentially exposed individuals need to be tested. Instead of pooling samples from all 15 together, how about first pooling 3 each to generate 5 sub-pools and then making one main pool from those 5 sub-pools? This way if the main pool turns up positive, instead of immediately resorting to testing all 15 individually, the next step could be to instead run the 5 sub-pools. If only one of those sub-pools turned up positive next, 12 of the 15 individuals would now automatically be deemed negative with no need to test them further. Sub-pooling in this instance would require a total of 9 tests to identify one positive among 15 instead of a total of 16 if only one master pool were used.
- Combinatorial or matrix pooling: Pool samples into groups such that each sample is part of multiple pools (4). Helps mitigate the risk of a false negative result from a low viral load sample when pooled with virus negative samples.
Learn from history: how does the blood supply industry use pooled samples to routinely screen blood and blood products for Hepatitis, HIV, etc.?
Test samples individually or pool them to detect an infectious agent isn’t a new question thrown up by the COVID-19 pandemic. Rather, it’s one that the blood supply system has had to struggle with ever since blood transfusions were found to transmit infectious diseases such as Hepatitis and HIV.
Blood donations and transfusions are now such a routine staple of life that the elaborate screening systems that exist to ensure that blood and blood products remain largely free of transmissible infectious agents remains quite invisible to the public.
Advocacy by famous examples such as Ryan White, a hemophiliac who contracted HIV through a contaminated blood donation, helped usher in the reactionary but preventive system that tries to minimize such risk (5).
However, prohibitive cost means that some of this screening is done on minipools, i.e., samples ranging from a few to many dozens pooled together prior to running the screening test to detect a given infectious agent (6). Sample pooling is also routine in the livestock industry and wildlife surveillance programs for detecting foot-and-mouth and other infections.
These fields have already worked out many of the general kinks such as sample size relationship to test sensitivity that could be directly applied to SARS-CoV-2 surveillance testing.
Illustrative evidence that pooled/group testing could be useful in SARS-CoV-2 detection
- A retrospective analysis using presumably clinical sample pooling found 2 positives among 2740 samples in the San Francisco Bay area using pools of 10 (7).
- A retrospective analysis using clinical sample pooling of either 5 or 10 detected positives as long as the unpooled sample had a viral load of at least a million RNA copies per swab (8). The study also suggested that targeting two rather than just one SARS-CoV-2 gene in RT-PCR tests might increase the efficiency of detection when pooling samples with low viral loads.
- A retrospective analysis using RNA extracts in Homburg, Germany, could detect 23 positives among 1191 using only 267 tests based on pool sizes ranging from 4 to 30 (9).
Ultimately, sample pooling for SARS-CoV-2 needs to prove its utility without compromising test sensitivity and specificity to provide results within the limits of detection of the test in order for its cost-effectiveness to be worth it. It will likely prove its worth in some places but not in others. Infection prevalence at the time of pooled sample testing might turn out to be the wild card in all of this; higher the prevalence, lower its utility.
Bibliography
1. Opinion | Five People. One Test. This Is How You Get There.
2. Dorfman, Robert. “The detection of defective members of large populations.” The Annals of Mathematical Statistics 14.4 (1943): 436-440. https://www.sis.uta.fi/tilasto/liski-arkisto/mtt-perusteet10/mttp-kurssi10/Materiaalia/Dorfman-Ann1943.pdf
3. Evaluation of Sample Pooling for Screening of SARS-CoV-2
4. Ben-Ami, Roni, et al. “Large-scale implementation of pooled RNA extraction and RT-PCR for SARS-CoV-2 detection.” Clinical Microbiology and Infection (2020). Large-scale implementation of pooled RNA extraction and RT-PCR for SARS-CoV-2 detection
6. Perkins, Herbert A., and Michael P. Busch. “Transfusion‐associated infections: 50 years of relentless challenges and remarkable progress.” Transfusion 50.10 (2010): 2080-2099.
7. Hogan, Catherine A., Malaya K. Sahoo, and Benjamin A. Pinsky. “Sample pooling as a strategy to detect community transmission of SARS-CoV-2.” Jama 323.19 (2020): 1967-1969. Sample Pooling as a Strategy to Detect Community Transmission of SARS-CoV-2
8. Torres, Ignacio, Eliseo Albert, and David Navarro. “Pooling of nasopharyngeal swab specimens for SARS‐CoV‐2 detection by RT‐PCR.” Journal of Medical Virology (2020). Pooling of nasopharyngeal swab specimens for SARS‐CoV‐2 detection by RT‐PCR
9. Lohse, Stefan, et al. “Pooling of samples for testing for SARS-CoV-2 in asymptomatic people.” The Lancet Infectious Diseases (2020). Pooling of samples for testing for SARS-CoV-2 in asymptomatic people