SARS-CoV-2 sensitivity limbo – How low can we go?

To the Editor,

Since the beginning of the pandemic, molecular methods such as real-time RT-PCR have been used as references for severe acute respiratory syndrome (SARS-CoV-2) detection. With unprecedented demands for SARS-CoV-2 testing, and difficulties acquiring NAAT supplies, clinical laboratories are challenged with providing timely results. Rapid diagnostic tests (RDTs) are simple, rapid, and portable technologies that offer a potential solution to increase the diagnostic testing capacity. Recently, some RDTs have become licensed under emergency use authorization for SARS-CoV-2 detection in the laboratory or point-of-care settings (Food and Drug Administration (FDA), 2020), but despite their high specificity, the applicability of RDTs has been hampered by poor clinical sensitivity, which often falls below the ideal target product profiles recommended by the World Health Organization (Dinnes et al., 2020, World Health Organization (WHO), 2020a, World Health Organization (WHO), 2020b). In a recent systematic review and meta-analysis (Dinnes et al., 2020), the average pooled sensitivity of antigen-based RDTs was 56%, and values as low as 11.7% have been reported (Nagura-Ikeda et al., 2020). In contrast, a recent study in this journal by Porte et al. (2020) described high clinical sensitivity of an antigen-based RDT at 93.9%. Given the wide variability in RDT sensitivity, careful consideration is needed on the generalizability and applicability of these findings.

Traditionally, SARS-CoV-2 detection methods strive to achieve the highest sensitivity possible (LeBlanc et al., 2020). From an individual diagnostic perspective, the decreased analytical sensitivity of RDTs (∼10⁵ copies/mL vs. ∼10³ copies/mL for NAATs) would likely only be relevant during a short period in the acute stage of illness, or late in disease (LeBlanc et al., 2020, Wölfel et al., 2020, Mina et al., 2020, Larremore et al., 2020). As individuals with resolving viral loads are less likely to be infective [i.e., high cycle thresholds (Ct values) in real-time RT-PCR], this situation may adequately be served by an RDT (Wölfel et al., 2020, Mina et al., 2020, Larremore et al., 2020). However, on a population level, the identification of individuals with early or late disease would allow for more complete contact tracing and potentially lead to more case finding and interventions.

In recent publications, an alternative strategy has been proposed that might overcome the poor sensitivity of RDTs by repeat testing of target populations over time, thereby increasing the probability of capturing individuals who fall into a period of high viral shedding (Mina et al., 2020, Larremore et al., 2020). To date, the feasibility of repeat testing using RDTs has been hampered by limitations such as scalability (with low throughput devices), human and material resource requirements, and acceptability of repeat collections with the authorized specimen types (e.g., nasopharyngeal swabs). Regardless of the challenges of RDT implementation, thorough validation is required with consideration for factors that are method-, virus-, host-, and context-dependent.

With the above considerations in mind, key parameters that remain to be defined for repeat testing using RDTs is the minimal acceptable value for sensitivity, the optimal testing frequency in target populations, and for which population or setting RDTs would be of best benefit. If the goal is case-finding and containment, the sensitivity of an assay would logically influence the testing frequency, and ideally, should be targeted to ensure the maximal and accurate detection of SARS-CoV-2 in the target population. However, if capacity or resource limitations defines a certain testing frequency, high sensitivity becomes increasingly relevant. As these parameters have yet to be defined for RDTs, we find ourselves in a “sensitivity limbo,” asking not only “how low can we go?” for accurate SARS-CoV-2 detection, but also “how low should we go?”

Conflict of interest

The authors declare that they have no conflicts of interest related to this manuscript.

Funding source

There was no funding for this manuscript.

Ethical approval

Institutional research ethics board approval was not required for this work.