Laboratory-based molecular tests in SARS-CoV-2 infection diagnosed using RT-PCR

Ingrid Arevalo-Rodriguez*, Miriam Mateos-Haro, Jacqueline Dinnes, Agustin Ciapponi, Clare Davenport, Diana Buitrago-Garcia, Tayeb Bennouna-Dalero, Marta Roqué-Figuls, Ann Van den Bruel, Karin Jasmin von Eije, Devy Emperador, Lotty Hooft, René Spijker, Mariska MG Leeflang, Yemisi Takwoingi, Jonathan J Deeks

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Background: Diagnosing patients with a SARS-CoV-2 infection played a critical role in managing the COVID-19 pandemic and remains a priority for the transition to long-term management of COVID-19. Initial shortages of extraction and RT-PCR reagents impaired the desired upscaling of testing in many countries, which led to the search for alternatives to extraction and RT-PCR testing. Reference standard methods for diagnosing the presence of SARS-CoV-2 infection rely primarily on real-time reverse transcription-polymerase chain reaction (RT-PCR). Alternatives to RT-PCR could, if sufficiently accurate, have a positive impact by expanding the range of diagnostic tools available for the timely identification of people infected by SARS-CoV-2 , access to testing and the use of resources.

Objectives: To assess the diagnostic accuracy of alternative (to RT-PCR assays) laboratory-based molecular tests for diagnosing SARS-CoV-2 infection.

Search methods: We searched the COVID‐19 Open Access Project living evidence database from the University of Bern until 30 September 2020 and the WHO COVID-19 Research Database until 31 October 2022. We did not apply language restrictions.

Selection criteria: We included studies of people with suspected or known SARS‐CoV‐2 infection, or where tests were used to screen for infection, and studies evaluating commercially developed laboratory-based molecular tests for the diagnosis of SARS-CoV-2 infection considered as alternatives to RT-PCR testing. We also included all reference standards to define the presence or absence of SARS‐CoV‐2, including RT‐PCR tests and established clinical diagnostic criteria.

Data collection and analysis: Two authors independently screened studies and resolved disagreements by discussing them with a third author. Two authors independently extracted data and assessed the risk of bias and applicability of the studies using the QUADAS‐2 tool. We presented sensitivity and specificity, with 95% confidence intervals (CIs), for each test using paired forest plots and summarised results using average sensitivity and specificity using a bivariate random‐effects meta‐analysis. We illustrated the findings per index test category and assay brand compared to the WHO's acceptable sensitivity and specificity threshold for diagnosing SARS-CoV-2 infection using nucleic acid tests.

Main results: We included data from 64 studies reporting 94 cohorts of participants and 105 index test evaluations, with 74753 samples and 7517 confirmed SARS-CoV-2 cases. We did not identify any published or preprint reports of accuracy for a considerable number of commercially produced NAAT assays. Ninety-two cohorts were judged at unclear or high risk of bias in more than three QUADAS-2 domains. Around half of the cohorts were considered at high risk of selection bias because of recruitment based on COVID status. Seventy-five out of 94 cohorts were at high risk of bias in the reference standard domain because of reliance on a single RT-PCR result to determine the absence of SARS-CoV-2 infection. Seventy cohorts were at unclear risk of bias due to a lack of clarity about the time interval between the index test assessment and the reference standard, the number of missing results, or the absence of a participant flow diagram.

For index tests categories with four or more evaluations and when summary estimations were possible, we found the following results: a) For RT-PCR assays omitting/adapting RNA extraction, the average sensitivity was 95.1% (95% CI 91.1% to 97.3%), and the average specificity was 99.7% (95% CI 98.5% to 99.9%; based on 27 evaluations, 2834 samples and 1178 SARS-CoV-2 cases); b) For RT-LAMP assays, the average sensitivity was 88.4% (95% CI 83.1% to 92.2%), and the average specificity was 99.7% (95% CI 98.7% to 99.9%; 24 evaluations, 29496 samples and 2255 SARS-CoV-2 cases); c) for TMA assays, the average sensitivity was 97.6% (95% CI 95.2% to 98.8%), and the average specificity was 99.4% (95% CI 94.9% to 99.9%; 14 evaluations, 2196 samples and 942 SARS-CoV-2 cases); d) for digital PCR assays, the average sensitivity was 98.5% (95% CI 95.2% to 99.5%), and the average specificity was 91.4% (95% CI 60.4% to 98.7%; five evaluations, 703 samples and 354SARS-CoV-2 cases); e) for RT-LAMP assays omitting/adapting RNA extraction, the average sensitivity was 73.1% (95% CI 58.4% to 84%), and the average specificity was 100% (95% CI 98% to 100%; 24 evaluations, 14342 samples and 1502 SARS-CoV-2 cases).

Only two index test categories fulfil the WHO-acceptable sensitivity and specificity requirements for SARS-CoV-2 nucleic acid tests: RT-PCR assays omitting/adapting RNA extraction and TMA assays. In addition, WHO-acceptable performance criteria were met in a meta-analysis of data for two assays out of 35 when tests were used according to manufacturer instructions. At 5% prevalence using a cohort of 1000 people suspected of SARS-CoV-2 infection, the positive predictive value of RT-PCR assays omitting/adapting RNA extraction/purification will be 94%, with three in 51 positive results being false positives, and around 2 missed cases. For TMA assays, the positive predictive value of RT-PCR assays will be 89%, with 6 in 55 positive results being false positives, and around one missed case.

Authors' conclusions: Alternative laboratory-based molecular tests aim to enhance testing capacity in different ways, such as reducing the time, steps and resources needed to obtain valid results. Several index test technologies with these potential advantages have not been evaluated or have been assessed by only a few studies of limited methodological quality, so the performance of these kits was undetermined.

Only two index test categories with enough evaluations for meta-analysis fulfil the WHO set of acceptable accuracy standards for SARS-CoV-2nucleic acid tests: RT-PCR assays omitting/adapting RNA extraction and TMA assays. These assays might prove to be suitable alternatives to RT-PCR for identifying people infected by SARS-CoV-2, especially when the alternative would be not having access to testing. However, these findings need to be interpreted and used with caution because of several limitations in the evidence, including reliance on retrospective samples without information about the symptom status of participants and the timing of assessment. Although we used a comprehensive search and had broad eligibility criteria to include a wide range of tests that could be alternatives to RT-PCR methods, further research is needed to assess the performance of alternative COVID-19 tests and their role in pandemic management.
Original languageEnglish
JournalCochrane Database of Systematic Reviews
Publication statusAccepted/In press - 14 Aug 2024

Bibliographical note

Not yet published as of 25/09/2024.

Cite this