Though the technique has seen a rise in popularity since the discovery of SARS‑CoV‑2, pooled testing has existed for decades and has far-reaching applications beyond the current pandemic. The outbreak of the COVID-19 virus in 2020 has exposed many supply chain and operational issues. Laboratories have not been exempt from this and have experienced a sharp increase in their needed testing output due to the rise in coronavirus tests to be analyzed. The speed of the virus, which according to the Centre for the Mathematical Modelling of Infectious Diseases, has an effective reproduction estimate of 1.2, left no room to prepare and insufficient time and funds to address laboratory constraints. With FDA guidance, many of these labs have turned to adopting a pooled testing technique to combat the backlog of samples. But what is pooled testing, and how does it impact the labs that use it?
Defining pooled testing:
Pooled testing refers to a class of experiments that follow the principle of testing a composite from a group of samples rather than testing each sample individually. While the exact group size and testing methodology can change based on the requirements of the lab, pooled testing is a way to avoid performing unnecessary experiments when the outcome of a sample result can be determined with logic and statistics. This procedure can often be found in the laboratory due to the nature of sample testing and has permeated other fields such as medicine, biological research, computer science, and mathematics.As with any operational strategy, there are benefits and tradeoffs towards adopting a pooled testing plan in the laboratory. Reducing the overall number of tests performed provides an immediate increase in both time and labor efficiency. In turn, this reduces the cost of analysis per sample batch and the amount of reagents needed to complete experiments. Effective implementation of a pooled testing strategy can also increase laboratory volume by enabling a higher testing output while holding resources constant. These results do come with a caveat, a sensitivity loss for any specific test could arise due to the combination of samples. Since the samples are pooled, any measurement would be indicative of the group rather than solely the individual. In testing cases searching for the presence of a certain marker or condition, this sensitivity is not needed as a negative result can only be achieved if every member of the sample group is negative. Testing scenarios that extend beyond the binary presence or absence of a substance can lead to complications with a pooled approach. This can be overcome with individual testing of flagged group samples, but analysis that requires sensitive measurement of every sample would not be suited for pooled testing.
Origin of pooled testing:
Formed by statistician Robert Dorfman, the idea of pooled testing came about due to the need to find healthy and capable soldiers for the US Army. In 1943, the United States was engaged in World War II, fighting in both the Pacific and European theaters. This global two front-war required over 16 million soldiers, creating a massive focus on enlistments. Bringing in recruits was important for ensuring army positions were properly staffed, but the army faced a public health epidemic of syphilis among current and potential troops. This infection could spread amongst the population and incapacitate soldiers needed for battle. Additionally, positive syphilis cases would need to be treated before enlisting, leading to a reduced number of recruits in exchange for protecting healthy troops and preventing further spread. Screening enlistees for this bacterial infection was a significant undertaking and encountered issues due to the speed of needed results and limitations in testing at the time.Dorfman solved these issues with the introduction of pooled testing. He proposed grouping soldiers’ samples to reduce the number of tests needed in military syphilis screening. In this early method, blood samples would be pooled according to the sample population size, and a representative sample of that pool was tested. Any pool that generated a defective result was then analyzed using traditional individual sample testing. Pools that yielded a negative result would clear all of the members of that sample pool. This method ensured that positive syphilis results were identified with accuracy while allowing for a reduced number of tests and maximum screening throughput.
Pooled method evolution:
Robert Dorfman’s initial technique of pooled testing helped support the war effort and remained the only pooling method until Andrew Sterrett’s contribution in 1957. Sterrett recognized inefficiencies in Dorfman’s analysis of defective sample pools. Depending on the size of the sample pool used, a single defective case could result in the need for more rigorous individual testing. In large pools, this could drastically increase the testing time and cost, especially if multiple pools were found to be defective. In this newly proposed method, pools that were identified to contain a defective sample were scheduled for individual testing. However, once the defective sample was identified, the remaining individual samples would be once again pooled together and tested as a composite. If this pooled test passed, then all of the samples in the pool would be cleared. This work drew attention to the possibilities of an improved way to quickly identify the defective test in a pool and, as a result, further reduce the number of tests needed overall.Pooled testing would go through several iterations as researchers sought to improve the method by increasing sensitivity and reducing required tests. The next advancement came in 1959 when Milton Sobel and Phyllis Groll published their paper describing new ways to optimize the experiment based on the probability of a defective result. With the development of computer science, binary search was introduced to pooled testing in 1972. This further improved the method by splitting identified defective pools in two and treating them as new pooled samples. This would repeat with the pool that generated a defective result until the defective sample was found.
Engaging with the possibility of mislabeling items led to the development of non-adaptive testing. In this method of pooled testing, rather than proceed with experiments based on the outcome of prior experiments, for example Dorfman’s method switching to individual testing upon a defective result, the entire experiment is planned and executed with no impact from the test results. The first of such models was the combinatorial orthogonal matching pursuit algorithm, or COMP. This concept was further iterated into the definite defectives, DD, and sequential COMP, SCOMP, algorithms that aimed to remove the issue of false positives. The development of combinatorial models such as Dorfman and Sterrett’s methodologies as well as non-adaptive strategies has created an array of pooled testing techniques that can be suited for laboratory needs.