Sensitivity and efficiency of RNA sample pooling for real-time quantitative polymerase chain reaction testing for SARS-CoV-2

Introduction

The health burden resulting from the outbreaks of COVID-19 in middle-low and low-income countries (ML-LIC) has been staggering. As of Nov 12 2020, ML-LIC had experienced 23.4% of all cases (about 11.5 million) and 22.0% of all deaths in the world [1,185,331] (1). Even though they has helped curtail the magnitude of the COVID-19 pandemic, most of the world population still have very limited access to effective vaccines, and COVID-19 is becoming an endemic disease in rich and poor countries (2-5). The need for booster shots, the emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants of concern, and the lack of accessible effective therapeutics contribute to the need of access to accurate testing in ML-LIC (6-8). Widespread access to COVID-19 testing to detect individuals with asymptomatic infections is key to contain the spread of the virus, decrease mortality, and avoid substantial economic losses in ML-LIC countries and poor populations (9).

Mass testing for SARS-CoV-2 infection makes feasible the identification and isolation of symptomatic and asymptomatic infectious individuals (10), and has been key in stopping the spread of the virus in many countries (11). Faced with limited testing capacity, ML-LIC countries have made limited use of real-time quantitative polymerase chain reaction (RT-qPCR) tests for preventing SARS-CoV-2 transmission (12). Indeed, official country level data (13) indicate the ratio of RT-qPCR tests per person has been 11 times higher in high income countries (6.7) than in ML-LIC countries (0.6), and the number of tests per confirmed case has been 38% higher (17.5 vs. 12.7, respectively). In order to save testing resources, several countries use a protocol based on the detection of the envelope (E) gene as a single test target for the diagnosis of SARS-CoV-2 infection (14-19).

To test as many people as possible, with limited resources, it is possible to conduct pooled tests for groups of people, rather than individuals, and conduct further testing in the individuals included in positive pools. Pool testing could be used as a screening approach for disease surveillance, expeditious studies of outbreaks, and in groups of people sharing the same environment. For instance, widespread testing of health care workers could be achieved by weekly testing using pool testing, regardless of their history of exposure to SARS-CoV-2, a task may not be feasible in resource limited setting (20,21).

Pool testing has been scarcely used for SARS-CoV-2 infection diagnosis in ML-LIC countries. This is in large part due to a lack of data on the sensitivity and efficiency of RT-qPCR pool testing, as compared to individual testing, particularly when the E gene is used as the single test target for diagnosis. We evaluated the sensitivity of RT-qPCR tests in RNA pools of 4, 8, 16, and 32 specimens, using the SARS-CoV-2 E gene as a single target, and assessed the relative efficiency of pool testing to detect true positive (TP) cases of infection. We show that pool testing is several times more efficient than individual testing to diagnose, identify, and isolate infected individuals, particularly if they are asymptomatic. We present the following article in accordance with the STARD reporting checklist (available at https://jphe.amegroups.com/article/view/10.21037/jphe-21-97/rc).

Methods

We used respiratory samples from the Central Research Laboratory (CRL), Universidad Industrial de Santander (UIS), Bucaramanga, Colombia, collected on July-September 2020, from symptomatic and asymptomatic individuals, being tested for clinical care or epidemiologic surveillance. The study was approved by the UIS’ and the University of Wisconsin-Madison’s Institutional Review Boards (IRB00003739). The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). No informed consent was required, since the study was based on respiratory samples stored before the start of our study.

We selected specimens of extracted RNA from SARS-CoV-2 RT-qPCR positive tests, independently of their cycle threshold (Ct). Single positive specimens were mixed with 3, 7, 15, and 31 negative specimens to make pools of 4, 8, 16, and 32 samples each. Pools of extracted RNA samples were used, instead of pre-extraction pools of nasopharyngeal swaps, because they had been stored and were readily available. Before including an originally positive sample in a pool, the sample was re-tested to confirm it was still positive and to get a new Ct value for that sample. The sample was included in a pool only if it was positive in this second individual test.

We tested at least 89 pools of each size, and ensured that about 25% of pools of each size had a positive specimen with a low viral load (Ct >34 cycles), to match the proportion of specimens with low viral load in the first 1,350 tests conducted in our lab. A Ct >34 cycles was pre-defined, because a dilution of a RT-qPCR positive sample mixed with 31 additional negative samples would result, on average, in an increase of five Ct cycles. This would increase the false negative rate for pools of 32 specimens, since the resulting Ct value would approach the cut-point of 40 Ct cycles.

RT-qPCR testing

Nasopharyngeal swabs and tracheal aspirates were collected and transported following a standard protocol (22). RNA was extracted using MagMAX^TM Viral/Pathogen II (MVP II) Nucleic Acid Isolation Kit (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer’s instructions. Purified RNA was stored at −80 ℃ in Micro Amp 96-well real-time optical PCR plates, until testing, 1 to 2 weeks later. A pool mix was prepared using 5 µL of purified RNA for final volumes of 20, 40, 80, and 160 µL for pools of size 4, 8, 16, and 32 samples, respectively. Five µL from the pool mix were used for RT-qPCR analysis, following the Berlin Protocol (23). A 25-µL reaction was set up containing 5 µL of RNA, 12.5 µL of 2× reaction buffer, 1 µL of reverse transcriptase/Taq mixture from Superscript III one-step RT-qPCR system (Invitrogen, Waltham, MA, USA), and 0.4 µL of a 50 mM magnesium sulfate solution. The test uses one primer and probe to detect a region in the SARS CoV-2 E gene at the concentration recommended by developers (23). Thermal cycling was performed at 55 ℃ for ten minutes for reverse transcription, followed by 95 ℃ for 3 min and then 45 cycles of 95 ℃ for 15 s, and 58 ℃ for 30 s. Positive and negative controls were included in each test. Tests with a Ct ≤40 cycles were considered positive.

Statistical analysis

We assumed viral RNA would have been detected in negative samples, if additional amplification cycles were used. Therefore, to prevent potential bias resulting from assigning a value of 40 Ct to these samples (24), we treated their Ct values as a right-censored variable, missing at random, conditionally on the Ct of the individual positive sample included in the pool. We used interval regression to obtain imputed values as low as 40 and as high as 42 Ct cycles, for negative samples, and generated 100 complete datasets (24,25).

We computed the average change in Ct value as the difference (Δ) between the individual Ct of the positive sample in the pool, the first time the individual sample was tested, when received in our lab, and the Ct of its pool, and its standard deviation (SD_Δ). We took a random value (Ct_r) of the distribution of Δ for each pool, normal (Δ, SD_Δ), and added it to the Ct value of each of the first 1,350 positive samples in our lab (Ct_0i) to obtain dilution-corrected Ct values. We calculated the sensitivity for each pool size (Sens_pool) as the proportion of the original 1,350 samples with a dilution-corrected Ct value (Ct_0i + Ct_r) >40 cycles (19). This process was repeated 1,000 times for each pool size, and the median and the upper and lower 2.5% of the simulated distribution were taken as Senspool and its 95% confidence interval (CI), respectively. This analysis was also conducted using the Ct value from the positive sample in the pool, tested at the time when the pool including that sample was also tested.

We wrote a program in Stata (StataCorp.; 2017; Stata Statistical Software: Release 15; College Station, TX: StataCorp LLC) to conduct simulations of pool testing by binary splitting and evaluate the efficiency of pool testing as compared to individual testing. We used Sens_pool for poll testing and (Sens_ind) of 95% (26,27) and specificity (Spec_ind) of 99.5% (26) for individual testing. For nested pools, we assumed a sensitivity of 100%. Efficiency was defined as the ratio of tests used per TP cases detected for individual (E_ind) and pool testing (E_pool) (28). And relative efficiency as (E_pool/E_ind)/3 (18,27,29). We took a third of the relative efficiency because we used pools of RNA, instead of pools of respiratory samples, and the amplification phase comprised a third of the total cost of a RT-qPCR test.

Results

We analyzed 388 SARS-CoV-2 positive and 5,696 negative specimens. Positive specimens had an average Ct of 26.6 cycles (95% CI: 25.8, 27.3) and 22% to 25% of all pools had one positive sample with a Ct >34 cycles (Table 1). The mean Ct in pooled samples was 29.6 cycles (95% CI: 28.9, 30.3), and was homogeneous across pools of different sizes. In contrast, the average change in Ct was 3.16 cycles (95% CI: 2.82, 3.50), and ranged from 2.21 in pools of 4 to 3.94 in pools of 32 samples.

The mean Ct value in the first 1,350 positive tests in our lab, used to estimate Senspool, was 27.9 cycles (95% CI: 27.5, 28.3), and 25.4% of them (95% CI: 22.9%, 28.0%) had a Ct >34 cycles. Sens_pool estimated from dilution-corrected Ct values, ranged from 82% in pools of size 32 to 91% in pools of size 4, and were similar, but considerably more precise, than to those based on the proportion of positive pools (Table 2).

The efficiency of individual testing increased with the prevalence of infection. For a doubling of the prevalence, efficacy increased by 100%: from 212, to 106, to 53 tests used per TP case detected, for prevalence of 0.5%, 1%, and 2%, respectively (Table 3). Pool testing efficacy also increased with prevalence, but the increase diminished with higher prevalence. For instance, for pools of 32 samples, increases in prevalence from 0.5% to 1% and 1% to 2% resulted in a 31% and 22% higher efficiency, respectively. Moreover, the smaller the pool size the higher the increase in efficiency associated to higher prevalence. Indeed, for pools of size 4, increases in prevalence from 0.5% to 1% and 1% to 5% improved efficiency by of 87% and 80%, respectively.

Although the relative gain in efficiency associated with higher prevalence was smaller for pool testing than from individual testing, pool testing was several times more efficient than individual testing in all scenarios, excepting pools of size 4. The efficiency of pools of size 8, 16, and 32 was 2.0 to 3.8, 1.7 to 2.5, and 1.3 to 1.5 times higher than that of individual testing, for prevalence of 0.5%, 1% and 2%, respectively. In contrast, the efficiency of pools of size 4 was similar to that of individual testing, regardless of the level of prevalence.

Using pool testing largely reduced the number of tests needed to detect the same number of TP cases that would have been detected using a fixed number of individual tests. For instance, if 1% of the population were infected, 20,000 tests would be needed to identify 190 TP cases, compared to about 2,600 tests if pools of size 16 were used (Table 3). In addition, pool testing resulted in a large increase in the number of individuals tested in a population, for a fixed number of available tests, regardless of baseline prevalence of infection. The larger the pool size, the higher the relative increase in the number of individuals tested. For instance, 20,000 tests used in pools of size four would allow testing 3.1, 3.4, and 3.8 times more people, for prevalence of 2%, 1%, and 0.5%, than individual testing. This allows detecting very large numbers of true negative, in addition to TP cases, by pool testing.

Findings from the analysis comparing the Ct from the test conducted in the positive sample when the sample was included in a pool and the Ct from the pool tests were virtually identically to those from the analysis using the original (first) Ct value for the positive sample.

Discussion

The average difference in Ct between individual and pooled tests in our study, from 2.21 in pools of size 4 to 3.94 in pools of size 32, were consistent with expected changes of approximately 1 Ct cycle for each twofold dilution of a positive sample (2 to 5 Ct) (30). They were also similar to those found in other studies using the E gene and RNA pools of similar sizes (18,19,31).

Pools of 4, 8, 16, and 32 specimens had sensitivity of 91.1%, 89.6%, 85.8%, and 82.5%, respectively. These estimates were similar, but more precise than those from Gupta et al. (95.4%, 95% CI: 77.1%, 99.9%; in pools of 8 samples) (31), Ben-Ami et al. (100%, 95% CI: 78.2%, 100%; in pools of 8 samples of nasopharyngeal swaps) (18), and Garg et al. (100%, 95% CI: 94.1%, 100%; in pools of ten samples of nasopharyngeal swaps) (32).

The efficiency of pools of eight or more samples was 30% to 280% higher than that of individual testing, depending on the level of prevalence. The relative efficiency of pools of size 32, without discounting costs of specimen collection and RNA extraction, were 11.5, 7.5, and 4.6 for prevalence of 0.5%, 1%, and 2%. These figures were similar to those obtained using de Wolff et al.’s simulation approach: 12.6, 8.6, and 3.6, respectively, had they not assumed a doubling of the time to do the pooled tests (29). Efficiency estimates for pools of size 4 were also similar to those from Aragón-Caqueo et al. (33). Moreover, pool testing efficiency increases significantly at sensitivity values much lower than those in our study (18,27,29). Indeed, for a sensitivity of 75% and a prevalence of 2%, pools of 32 samples would have a relative efficiency of 4.5, according to our simulations.

In our simulations, we assumed sensitivity was 100% in pools nested within a parent positive pool. If a nested pool were negative, most labs would continue testing the smaller nested pools, because it is safer assuming the parent pool test was a TP than a false positive (FP) (18). Were the parent pool test a FP, further testing in nested pools would identify it as such, because repeated testing in nested pools practically decreases the FP rate to zero (29). This characteristic makes pool testing particularly useful for mass testing of populations with low prevalence of infection. In contrast, individual testing could result in higher numbers of FP than TP, if the FP rate is higher than the prevalence of infection.

Due to its high efficiency and extremely low FP rate, pool testing makes possible not only controlling outbreaks, but also loosening lock-down and social distancing restrictions, by preventing the isolation of FP cases. For a 1% prevalence of infection, using 20,000 RT-qPCR tests in RNA pools of 32 samples would allow discarding SARS-CoV-2 infection in more than 180,000 individuals, who could safely return to work. Moreover, if the main goal is identifying uninfected individuals and returning them to work, pool testing could be used in populations with a high point prevalence of SARS-CoV-2 infection, such as health care workers.

A narrow focus on patient care has prevented the use of pool testing in ML-LIC. This has hampered the detection, tracking, and isolation of asymptomatic cases to stop the transmission of SARS-CoV-2. Even when the fraction of positive tests and the demand for tests have decreased, efforts for mass testing have been limited. Although mass testing initially increases the number of cases, in the long term the strategy leads to lower viral transmission, decreased mortality, and quicker reactivation of the economy (34).

Prevention campaigns based on testing symptomatic and high risk individuals (35), likely have a limited impact on prevention, since asymptomatic infections, which occur in 31% of all cases (36), are responsible for a large fraction of new infections (36). These findings provide strong support for universal instead of selective testing, as a better strategy to prevent new infections, and mitigate health and economic losses (34).

In spite of the abundance of evidence in support of pool testing, from proof of concept (19), statistical (37), simulation (29), and field studies (31,32), recommendations from international health agencies have been non-committal and confusing (17,38). Indeed, it is well established that pool testing could increase cost-efficiency even in populations with a prevalence of infection up to 38% (39), if two-stage Dorfman pooling is used, and up to 10% if binary splitting by halving is used (40). Given the short duration of COVID-19, the prevalence of SARS-CoV-2 infection on the day of testing would be low in the general population, which is the target for mass testing (41,42). In Uruguay, the only LA country using pool testing to prevent transmission (43), the number of tests per case detected (75.2) is 53.4 times higher than in the rest of the region (1.4), and the number of deaths per million is 59 times lower (18.4 vs. 1,064.7) (13).

Ideally, the sensitivity of pool testing should be estimated in each lab, to account for differences in methods and experience. However, the Δ and SD_Δ from this and other studies could be used to estimate lab-specific dilution-corrected Ct values and sensitivity of pooled tests in other labs. It is also key for each lab to quantify the additional time required for preparing pools and for positive pools deconvolution to identify positive individuals. Additional time over single testing would depend on each lab specific requirements and resources, poll size used, and prevalence of active infection. This time should be taken into account while calculating the local cost-efficiency of pool testing. There is scarce data on pool testing as compare to individual testing, since this varies significant from lab to lab. However, findings from simulation studies in large populations suggest that pool testing for population-wide screening, such as in health care workers and essential personnel, could be 8–10 times faster than individual testing (29,44). Labs could also use matrix pool testing, i.e., two-dimensional array of rows and columns, with each sample included in a row-pool and a column pool to avoid pool deconvolution (45,46). If both pools defining an intersecting cell in a two-dimensional array were positive, then the shared sample in that cell would be positive, and no additional testing would be needed (29). We have adopted this approach in our lab.

Our estimates of sensitivity are conservative due to the use of a single gene target. Even though the E gene is the most sensitive to detect SARS-CoV-2 (23,47). Moreover, our pooled test sensitivity and efficiency could be underestimated, because our simulation approach did not account for clustering of positive samples in the same pool (42). Another limitation of our study is the use of pools of RNA, which were readily available, instead of pools of nasopharyngeal swaps before RNA extraction. The latter approach saves time and resources dedicated to RNA extraction in single samples. Therefore, our estimates of pool testing efficiency are lower than they would be if pools of nasopharyngeal swaps were used.

In summary, pool testing could increase the number of people tested several times, decrease the cost of testing per TP detected, decrease test result reporting time, and curtail the transmission of SARS-CoV-2, by making possible the identification and isolation of symptomatic and asymptomatic infected individuals.

Acknowledgments

Funding: This work was supported by the Mayor’s Office, City of Bucaramanga, Colombia and the Cámara Colombiana de la Construcción, Regional Bucaramanga (CAMACOL).

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://jphe.amegroups.com/article/view/10.21037/jphe-21-97/rc

Data Sharing Statement: Available at https://jphe.amegroups.com/article/view/10.21037/jphe-21-97/dss

Peer Review File: Available at https://jphe.amegroups.com/article/view/10.21037/jphe-21-97/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jphe.amegroups.com/article/view/10.21037/jphe-21-97/coif). LEB serves as an unpaid editorial board member of Journal of Public Health and Emergency from February 2020 to January 2024. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by the UIS’ and the University of Wisconsin-Madison’s Institutional Review Boards (IRB00003739). The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). No informed consent was required, since the study was based on respiratory samples stored before the start of our study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Coronavirus pandemic (COVID-19). 2020. Available online: https://ourworldindata.org/coronavirus
2. Burki T. Global COVID-19 vaccine inequity. Lancet Infect Dis 2021;21:922-3. [Crossref] [PubMed]
3. Figueroa JP, Hotez PJ, Batista C, et al. Achieving global equity for COVID-19 vaccines: stronger international partnerships and greater advocacy and solidarity are needed. PLoS Med 2021;18:e1003772. [Crossref] [PubMed]
4. Telenti A, Arvin A, Corey L, et al. After the pandemic: perspectives on the future trajectory of COVID-19. Nature 2021;596:495-504. [Crossref] [PubMed]
5. Antia R, Halloran ME. Transition to endemicity: understanding COVID-19. Immunity 2021;54:2172-6. [Crossref] [PubMed]
6. Connor A, Hariharan N, Carson S, et al. Access to COVID-19 testing in low- and middle-income countries is still critical to achieving health equity. Health Affairs Blog 2021. doi: 10.1377/hblog20211026.483412.
7. Mendoza AJM, Dela Cruz MDM, Gozum IEA, et al. The need for mass testing and the proper post-COVID-19 test behavior in the Philippines. J Public Health (Oxf) 2021;43:e348-9. [Crossref] [PubMed]
8. Rosenthal PJ. The importance of diagnostic testing during a viral pandemic: early lessons from novel coronavirus disease (COVID-19). Am J Trop Med Hyg 2020;102:915-6. [Crossref] [PubMed]
9. Dizioli A, Andrle M, Bluedorn J. COVID-19 hits the poor harder, but scaled-up testing can help. 2020. Available online: https://blogs.imf.org/2020/12/03/covid-19-hits-the-poor-harder-but-scaled-up-testing-can-help/
10. Walensky RP, Del Rio C. From mitigation to containment of the COVID-19 pandemic: putting the SARS-CoV-2 genie back in the bottle. JAMA 2020;323:1889-90. [Crossref] [PubMed]
11. Kennedy DS, Vu VK, Ritchie H, et al. COVID-19: identifying countries with indicators of success in responding to the outbreak. Gates Open Res 2021;4:62. [Crossref] [PubMed]
12. Hernández-Vásquez A, Barrenechea-Pulache A, Azañedo D. COVID-19 testing in Peru: low access and inequalities. J Public Health (Oxf) 2021;43:e504-6. [Crossref] [PubMed]
13. Hasell J, Mathieu E, Beltekian D, et al. A cross-country database of COVID-19 testing. Sci Data 2020;7:345. [Crossref] [PubMed]
14. Ramírez JD, Muñoz M, Hernández C, et al. Genetic diversity among SARS-CoV2 strains in South America may impact performance of molecular detection. Pathogens 2020;9:580. [Crossref] [PubMed]
15. Praharaj I, Jain A, Singh M, et al. Pooled testing for COVID-19 diagnosis by real-time RT-PCR: a multi-site comparative evaluation of 5- & 10-sample pooling. Indian J Med Res 2020;152:88-94. [Crossref] [PubMed]
16. Agoti CN, Mutunga M, Lambisia AW, et al. Pooled testing conserves SARS-CoV-2 laboratory resources and improves test turn-around time: experience on the Kenyan Coast. Wellcome Open Res 2021;5:186. [Crossref] [PubMed]
17. Organización Panamericana de la Salud. Directrices de laboratorio para la detección y el diagnóstico de la infección con el virus COVID-19. 2020. Available online: https://iris.paho.org/bitstream/handle/10665.2/52471/OPSIMSPHECOVID-19200038_spa.pdf?sequence=1&isAllowed=y
18. Ben-Ami R, Klochendler A, Seidel M, et al. Large-scale implementation of pooled RNA extraction and RT-PCR for SARS-CoV-2 detection. Clin Microbiol Infect 2020;26:1248-53. [Crossref] [PubMed]
19. Yelin I, Aharony N, Tamar ES, et al. Evaluation of COVID-19 RT-qPCR test in multi sample pools. Clin Infect Dis 2020;71:2073-8. [Crossref] [PubMed]
20. Rivara FP, Fihn SD, Perencevich EN. JAMA network open and COVID-19. JAMA 2020;324:1153-4. [Crossref] [PubMed]
21. Cherif A, Grobe N, Wang X, et al. Simulation of pool testing to identify patients with coronavirus disease 2019 under conditions of limited test availability. JAMA Netw Open 2020;3:e2013075. [Crossref] [PubMed]
22. Instituto Nacional de Salud. Lineamientos para la vigilancia por laboratorio de virus respiratorios. Bogotá: Minsalud, 2020. Available online: https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwiao6nNwP_1AhUhkokEHSe9DoEQFnoECAgQAQ&url=https%3A%2F%2Fwww.minsalud.gov.co%2Fsites%2Frid%2FLists%2FBibliotecaDigital%2FRIDE%2FVS%2FED%2FVSP%2Fpsps04-lineamiento-uso-pruebas-lsp-covid-19.pdf&usg=AOvVaw15W_2Qa9HuHP6LeRctgBk2
23. Corman VM, Eckerle I, Bleicker T, et al. Detection of a novel human coronavirus by real-time reverse-transcription polymerase chain reaction. Euro Surveill 2012;17:20285. [Crossref] [PubMed]
24. McCall MN, McMurray HR, Land H, et al. On non-detects in qPCR data. Bioinformatics 2014;30:2310-6. [Crossref] [PubMed]
25. Sterne JA, White IR, Carlin JB, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ 2009;338:b2393. [Crossref] [PubMed]
26. Afzal A. Molecular diagnostic technologies for COVID-19: Limitations and challenges. J Adv Res 2020;26:149-59. [Crossref] [PubMed]
27. Lohse S, Pfuhl T, Berkó-Göttel B, et al. Pooling of samples for testing for SARS-CoV-2 in asymptomatic people. Lancet Infect Dis 2020;20:1231-2. [Crossref] [PubMed]
28. Kim HY, Hudgens MG, Dreyfuss JM, et al. Comparison of group testing algorithms for case identification in the presence of test error. Biometrics 2007;63:1152-63. [Crossref] [PubMed]
29. de Wolff T, Pflüger D, Rehme M, et al. Evaluation of pool-based testing approaches to enable population-wide screening for COVID-19. PLoS One 2020;15:e0243692. [Crossref] [PubMed]
30. Gibson UE, Heid CA, Williams PM. A novel method for real time quantitative RT-PCR. Genome Res 1996;6:995-1001. [Crossref] [PubMed]
31. Gupta E, Padhi A, Khodare A, et al. Pooled RNA sample reverse transcriptase real time PCR assay for SARS CoV-2 infection: a reliable, faster and economical method. PLoS One 2020;15:e0236859. [Crossref] [PubMed]
32. Garg J, Singh V, Pandey P, et al. Evaluation of sample pooling for diagnosis of COVID-19 by real time-PCR: a resource-saving combat strategy. J Med Virol 2021;93:1526-31. [Crossref] [PubMed]
33. Aragón-Caqueo D, Fernández-Salinas J, Laroze D. Optimization of group size in pool testing strategy for SARS-CoV-2: a simple mathematical model. J Med Virol 2020;92:1988-94. [Crossref] [PubMed]
34. Cherif R, Hasanov F. A TIP against the COVID-19 pandemic. Washington: International Monetary Fund, 2020. Available online: https://books.google.com/books?hl=en&lr=&id=crEaEAAAQBAJ&oi=fnd&pg=PA4&dq=A+TIP+against+the+COVID-19++pandemic&ots=erXVoYEIP0&sig=pbIcntSczZxHkiDK_J1vxGVqNaA
35. Ministerio de Salud y Protección Social de Colombia. Manual de implementación PRASS: pruebas, rastreo y aislamiento selectivo sostenible. Bogotá: Minsalud, 2020. Available online: https://www.misgsst.com/pagina/Manual-de-implementacion-PRASS--Pruebas--Rastreo-y-Aislamiento-Selectivo-Sostenible
36. Buitrago-Garcia D, Egli-Gany D, Counotte MJ, et al. Occurrence and transmission potential of asymptomatic and presymptomatic SARS-CoV-2 infections: a living systematic review and meta-analysis. PLoS Med 2020;17:e1003346. [Crossref] [PubMed]
37. Armendáriz I, Ferrari PA, Fraiman D, et al. Group testing with nested pools. IEEE Trans Inf Theory 2021;68:1119-32. [Crossref]
38. World Health Organization. Diagnostic testing for SARS-CoV-2: interim guidance. Geneva: World Health Organization, 2020. Available online: https://www.who.int/publications/i/item/diagnostic-testing-for-sars-cov-2
39. Ungar P. The cutoff point for group testing. Communications on Pure and Applied Mathematics 1960;13:49-54. [Crossref]
40. Ben-Amotz D. Optimally pooled viral testing. Epidemics 2020;33:100413. [Crossref] [PubMed]
41. Lavezzo E, Franchin E, Ciavarella C, et al. Suppression of a SARS-CoV-2 outbreak in the Italian municipality of Vo'. Nature 2020;584:425-9. Erratum in: Nature 2021;590:E11. [Crossref] [PubMed]
42. Barak N, Ben-Ami R, Sido T, et al. Lessons from applied large-scale pooling of 133,816 SARS-CoV-2 RT-PCR tests. Sci Transl Med 2021;13:eabf2823. [Crossref] [PubMed]
43. Taylor L. Uruguay is winning against covid-19. This is how. BMJ 2020;370:m3575. [Crossref] [PubMed]
44. Verdun CM, Fuchs T, Harar P, et al. Group testing for SARS-CoV-2 allows for up to 10-fold efficiency increase across realistic scenarios and testing strategies. Front Public Health 2021;9:583377. [Crossref] [PubMed]
45. McMahan CS, Tebbs JM, Bilder CR. Two-dimensional informative array testing. Biometrics 2012;68:793-804. [Crossref] [PubMed]
46. Phatarfod RM, Sudbury A. The use of a square array scheme in blood testing. Stat Med 1994;13:2337-43. [Crossref] [PubMed]
47. Nalla AK, Casto AM, Huang MW, et al. Comparative performance of SARS-CoV-2 detection assays using seven different primer-probe sets and one assay kit. J Clin Microbiol 2020;58:e00557-20. [Crossref] [PubMed]