Opportunities and challenges in the use of pool sampling for sexually transmitted infections

Screening for sexually transmitted infections (STIs) is a fundamental strategy in preventing their spread, particularly for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG), which are the most prevalent and treatable bacterial STIs. Both can infect multiple anatomical sites, including urogenital, pharyngeal and anorectal sites in males and females. Nucleic acid amplification tests (NAATs) are currently considered the most reliable method for identifying these pathogens due to their high sensitivity and specificity in both urogenital and extragenital specimens. Current guidelines recommend testing samples individually from all three anatomical sites.¹ However, pooled testing offers a cost-effective and resource-efficient alternative by combining specimens from the three anatomical sites of a single individual into one test. This approach reduces the need for extensive sampling and testing resources, potentially lowering screening costs. Additionally, NAATs platforms have a limited capacity for daily tests, making it challenging busy laboratories to perform three separate tests for each patient. In September 2020, the CDC highlighted shortages of STIs test kits and laboratory supplies due to the COVID-19 pandemic.² Pooled testing could facilitate ongoing screening across all three sites, even when there are constraints on testing supplies or laboratory capacity. However, this strategy is not new. It was first proposed by economist Robert Dorfman in the 1940s to screen soldiers for syphilis.³ His work is the first published study on “pooled testing” and was notably applied in the field of STIs.

In the 1980s, pooled testing was subsequently used for population screening for HIV.⁴ Later, it was employed for the screening of various infections, before gaining more prominence during the SARS-CoV-2 pandemic.⁵ Pooled testing increases the number of people tested within the same budget, but cost savings depend on the pathogen's prevalence and pool size. Various formulas estimate the likelihood of a pool testing positive based on disease prevalence and pool size.^3,5,6 Dorfman's algorithm involves an initial round of group tests followed by individual tests for positive groups. The Mutesa algorithm uses a similar first round but with larger groups, followed by ‘slice tests’ to identify infected individuals, occasionally requiring additional rounds.⁵ Peeling et al.’s algorithm calculates the optimal pool size using the formula 1 − (1 − P)ⁿ. where P is the probability of a positive sample and n is the pool size.⁶ Non-adaptive algorithms, requiring only one round of testing, exist but have higher failure rates. Dorfman's method is commonly used in microbiology laboratories.³

From a practical standpoint, it is essential to differentiate between “pool testing” involving the mixing of samples from different patients and pool testing involving different anatomical samples from the same patient, as is the case with STIs. The latter approach is currently applied to STIs screening. When pooling samples, it is crucial to consider the pool size (for COVID-19, pool sizes typically range from 30 to 100 samples, with optimal efficiency at a 1% prevalence rate). However, for STIs, pool sizes usually consist of three anatomical samples. Additionally, when using NAAT as a diagnostic method, it is important to consider the limit of detection (expressed in inclusion-forming units/mL and colony-forming units/mL), the cycle threshold (Ct) value (especially for discordant results with low or high bacterial loads) to avoid false negatives, and whether the test is used for symptomatic or asymptomatic patients. From a quality perspective, it is also desirable for laboratories performing these pooled tests to be accredited.

This pool sampling approach raises several questions regarding its current implementation in microbiology laboratories.

The first issue is whether it is necessary to include extragenital samples in the screening of patients. Numerous studies have shown that excluding these samples can lead to a loss of positive cases, particularly in extragenital areas where asymptomatic patients have a higher prevalence rate. Verougstraete et al. evaluated the feasibility of molecular testing for CT and NG in pooled samples compared to individual samples from the pharynx, vagina, and rectum in a large group of female sex workers (FSW). They found that testing only vaginal samples would have led to missing 40% of CT infections and 60% of NG infections.⁷ Badman et al. further evaluated the performance of pooled urine, pharyngeal, and rectal specimens compared to individual anatomical specimens using the NAAT assay in men who have sex with men (MSM). They found that if only urine specimens had been tested, 82.0% (41 out of 50) of CT infections and 84.6% (33 out of 39) of NG infections would have been missed. Including extragenital samples, in CT and NG screening increases the detection rate of infections.⁸ Groups such as MSM and FSW, who have higher rates of extragenital infections, necessitate testing across three anatomic sites.⁹ Although this approach incurs additional costs, pooling genital and extragenital samples is a viable alternative. However, in low-prevalence populations, identifying the specific infected anatomical site is crucial as it can affect treatment decisions.¹

Various studies have been conducted to evaluate the use of pooled samples versus individual samples for detecting primarily CT and NG using different commercial systems, as reviewed mainly by Prazuck et al. showing a positive percentage agreement varying from 82.4% to 98.3% for NG, from 77.8% to 96.0% for CT, and 92.3% for MG.^7,10,11 A crucial factor in sample pooling is the volume used, as dilution can significantly affect results. Pooling dilutes bacterial load and can cause false negatives. Previous research indicates that reduced pooled sensitivity may be due from urine volume dilution, especially in single-site infections with low bacterial loads. Wilson et al. support this, noting higher pooled sensitivities in females (no urine added) compared to MSM. False-negative pooled specimens from MSM often had single-site infections, likely with low bacterial loads that fell below detection thresholds due to dilution. To enhance detection accuracy, studies involving MSM sample pooling should aim to reduce the volume of first-catch urine or consider using alternative urethral samples, such as meatal swabs, to mitigate the impact of dilution.^12,13

The third question addresses resource use in pooled testing. Cost savings from pooled testing depend on pool size and the prevalence of STIs in the population. Pooled testing is generally more cost-effective in low-risk populations, where the need for retesting positive pools is lower. In populations with higher STIs prevalence or higher rates of extragenital infections, such as MSM and some FSW, the cost-effectiveness of pooled testing decreases due to the increased need for retesting.

However, there is no consensus on the optimal pool size, and cost savings vary across studies. Some studies have shown cost savings of up to 85% by pooling vaginal swabs in groups of three in low-prevalence populations, such as high school female students with a 5.6% CT prevalence rate, and a 70% reduction for individual diagnosis after retesting positive pools.¹⁴ Verougstraete et al. demonstrated a 35% reduction in reagent costs and lab technician time by using pooled testing from three anatomic sites.⁷ Pooling urine samples in groups of three reduced reagent costs by 33%.¹⁵ Research by Wilson et al. demonstrated that pooling samples could reduce costs by up to £18.22 per person tested. This represents a substantial savings when scaled to the population level.¹²

Another question that arises for us is how to interpret the results after a positive pooled test. The study by Narvaez S et al. evaluated the efficacy of a rapid PCR test using a combined pool of urine, rectum, and pharynx samples from asymptomatic individuals engaged in risky sexual practices. Compared to standard PCR, the pooled method showed 77 (17.9%) concordant positive samples for CT and/or NG. Most discrepancies occurred in pharyngeal samples for NG, with 16 out of 426 (3.75%) discrepant results.¹⁶ These inconsistencies could be attributed to a lower bacterial load in the pharynx, Bissessor et al. propose that the bacterial load of NG plays a crucial role in transmission, their data reveals that the median bacterial load for rectal infections is notably higher than that for pharyngeal infections. This suggests that transmission from the rectum to the urethra may be more probable per sexual exposure compared to transmission from the pharynx, particularly in instances of symptomatic rectal infections.¹⁷

On the other hand, supplementary testing for confirmation of NG NAATs is still recommended for all samples tested, particularly those from extragenital sites.¹⁸ This is because it may also lead to false-positive results due to cross-reactions with commensal Neisseria sp. or Neisseria meningitidis, which have high nucleic acid homology to NG and may cross-react in the NAAT assay used. Then, after a positive test, might it be necessary to assess the bacterial load?

In CT, due the high specificity of properly validated NAATs, the relatively high prevalence in most European contexts, and the risk of missing low-positive results in repeated testing, confirmatory testing of NAAT-positive specimens is not advised [Low certainty; Grade 2].¹⁹

Multisite pooled testing presents a few potential limitations. Some studies have found a decrease in sensitivity.^7,10,11 Programs and services must evaluate whether this slight decrease in sensitivity is justified by the significant cost savings and potential improvements in health equity. Another potential limitation is that the oropharynx may have a lower bacterial load compared to genital and anorectal sites, which might lead to pooled testing missing oropharyngeal infections.¹⁸ Further research is required to enhance the detection of these infections. Moreover, it does not provide precise anatomical site data, which complicates treatment with specific antibiotics, as the choice of antibiotic can vary depending on whether the infection is urogenital, anorectal, or oropharyngeal. Although some guidelines recommend different treatments for CT and NG based on the infection site, the current WHO guidelines suggest that the first-line treatment can be consistent across infections, regardless of the anatomic site.²⁰ It could also impede the gathering of epidemiological data for reporting CT/NG infections, since the precise location of the infection would remain undetermined without individually testing each site. Decisions about STIs test selection should consider performance, costs, available resources, clinical benefits, and the acceptability of the testing method to both patients and clinicians.

The cost-effectiveness of pooled testing depends on the prevalence of STIs.¹⁴ In low-prevalence settings, pooled testing can increase the number of individuals tested and the frequency of testing for high-risk groups, such as those taking PrEP. This approach enhances access to STI testing and aligns with national guidelines recommending routine triple-site testing. Pooled testing improves the detection of CT/NG infections by including anorectal and pharyngeal specimens, which are often missed in single-site testing due to the higher prevalence of asymptomatic extragenital infections.^8,9

On the other hand, the acceptability of self-collecting samples can further reduce barriers to STI testing. By ensuring multisite testing among relevant populations, pooling can expand testing coverage and increase testing frequency for those at higher risk.

There are additional practical considerations for implementing pooled testing. First, clinic workflow must be considered, including the allocation of time for self-sampling. Second, laboratories must have the capacity to promptly retest individual samples from positive pools to avoid delays in treatment. Minimizing the interval between screening and treatment is crucial for reducing negative health outcomes and limiting further transmission. Quality control measures in laboratories should ensure efficient sample transport, minimize contamination risks, the amount of diluent used, provide adequate staff training, and maintain appropriate facilities for storing individual samples that may require retesting.

In conclusion, multisite pooled testing for STIs is a sensitive and specific method that has been adopted in specific contexts and appears suitable for the current volume of STI samples. This approach can enhance access to testing for more individuals and facilitate regular testing for extragenital site infections in relevant populations. However, it also presents limitations and challenges that need to be addressed in future research.