key: cord-0841520-sqmcwe3u
authors: Lee, Jonghoo; Song, Jae‐Uk
title: Diagnostic accuracy of the Cepheid Xpert Xpress and the Abbott ID NOW assay for rapid detection of SARS‐CoV‐2: A systematic review and meta‐analysis
date: 2021-05-03
journal: J Med Virol
DOI: 10.1002/jmv.26994
sha: 4e07f41b6a08fec25024419cdf9d95667396a24a
doc_id: 841520
cord_uid: sqmcwe3u

Rapid and accurate diagnosis of severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection is essential to prevent the spread of the virus. We investigated the diagnostic accuracy of the Xpert Xpress and the ID NOW assays for rapid detection of SARS‐CoV‐2 using a systemic review and meta‐analysis approach. A systematic literature search was performed using PubMed, Embase, and the Cochrane COVID‐19 Study Register. The sensitivity and specificity of these tests for detecting viruses in patients with suspected SARS‐CoV‐2 infection were pooled. We used commercial and laboratory‐developed reverse transcription‐polymerase chain reactions as reference standards. The Quality Assessment of Diagnostic Accuracy Studies‐2 tool was used to assess the risk of bias. We identified 11 studies involving 1734 subjects for the Xpert Xpress assay and 10 studies involving 1778 subjects for the ID NOW assay. The pooled sensitivity and specificity of the Xpert Xpress assay for detection of SARS‐CoV‐2 were 0.99 (95% confidence interval [CI], 0.97 to 0.99) and 0.97 (95% CI, 0.95 to 0.98), respectively. The pooled sensitivity and specificity of the ID NOW assay were 0.79 (95% CI, 0.69 to 0.86) and 1.00 (95% CI, 0.98 to 1.00), respectively. The studies included in our analysis seemed to have low methodological quality. The Xpert Xpress assay showed excellent diagnostic accuracy for rapid detection of SARS‐CoV‐2. However, as the ID NOW assay showed relatively low sensitivity, this test might miss several positive samples.

Coronavirus disease 2019 , caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has generated global health concerns since December 2019. 1, 2 Rapid and accurate diagnosis of SARS-CoV-2 infection can reduce the risk of virus transmission through early patient isolation and rapid institution of appropriate personal protective equipment use. 3 Nucleic acid amplification tests (NAATs) are considered the gold standard diagnostic method for detection of SARS-CoV-2. 4 Available NAATs commonly require batch testing, are time-consuming, and require a high degree of technical expertise. There also are commercially available rapid diagnostic NAATs that can be performed on demand, providing test results in <1 h.

The following rapid molecular tests have been currently ap- CoV-2 assay is a commercially available molecular test that detects the viral envelope E gene and the nucleocapsid N2 gene as its SARS-CoV-2-specific targets. 6 This platform integrates specimen processing, nucleic acid extraction, reverse transcription-polymerase chain reaction (RT-PCR) amplification of RNA, and amplicon detection using a single cartridge. 6 This assay has a short turnaround time of approximately 45 min. 6 The ID NOW COVID-19 assay is a rapid diagnostic test that utilizes isothermal amplification and can report results in <15 min. 7 Currently, the clinical utility of the two assays remains unclear.

Our objective was to evaluate the diagnostic performances of the Xpert Xpress and the ID NOW assays in rapid diagnosis of the SARS-CoV-2 virus from respiratory tract specimens. This evaluation was conducted through a systematic review and meta-analysis of clinical trial data.

This meta-analysis is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses of Diagnostic Test Accuracy Studies statement. 8 We performed a comprehensive search of three electronic databases (PubMed, Embase, and the Cochrane Central Register) ending in January, 2021. Search terms were "COVID-19 diagnostic testing" [MeSh] OR "Xpert" OR "Gene Xpert" OR "Xpert Xpress" OR "Cepheid" OR "ID NOW" OR "Abbott ID NOW" AND "COVID-19" OR "SARS-CoV-2" OR "2019-nCoV." The full search strategy for each database is provided in the Supporting Information. As this study was a systematic review of published articles, neither informed consent nor ethics approval was required. We also conducted a manual search of the references listed in relevant review articles.

We included studies that met the following inclusion criteria: (1) fulllength reports published in peer-reviewed English language journals;

(2) evaluations of the performance of the Xpert Xpress assay or the ID NOW assay compared with a reference standard; (3) inclusion of patients with SARS-CoV-2 infection; and (4) provision of sufficient data to calculate absolute numbers of true-positive, false-positive, false-negative, and true-negative results. Review articles, case reports, commentaries, and studies reporting outcomes without raw data or peer review were excluded. We also excluded preprint papers that did not receive the corresponding peer review. Participant demographics and underlying diseases were not restricted.

Respiratory specimens comprised nasopharyngeal (NP) aspirates, swabs, or washes; nasal aspirates, swabs, or washes; and throat swabs. The reference standard was either a commercial or laboratory-developed RT-PCR.

The two authors independently performed extractions of potentially relevant studies and reviewed each study according to predefined eligibility criteria, after which data were extracted. Any disagreements that arose during study selection or data extraction were resolved by discussion. A predefined form was used to extract data from each study.

The extracted data from each study included in the meta-analysis were author, study design, place of study, number of samples, age of subjects, sex of subjects, index test, comparison test, and type of specimens. As recommended by the Cochrane Collaboration, we used the Quality Assessment of Diagnostic Accuracy Studies (QUADAS)-2 tool to assess the risk of bias in diagnostic test accuracy. 9 The studies were considered to have a "low" risk of bias if the risk assessment was scored as "low" for patient selection, index test, reference standard, and flow and timing. If any domain was scored as a "high" risk of bias, or if two or more domains were assessed to have "unclear" bias, then the study was judged as having a "high" risk of bias. If a study was assessed as "unclear" in one of the four domains, the risk of bias was ranked as "unclear." 9 Discrepancies were resolved by consensus between the two authors.

For the diagnostic meta-analysis, the bivariate random-effects model was used for analysis, and diagnostic performance measures were pooled across studies. 10 The bivariate model estimates pairs of logit-transformed sensitivity and specificity from studies, incorporating the correlation that can exist between sensitivity and specificity. 10 We extracted the numbers of patients with true-positive, false-positive, false-negative, and true-negative test results either directly or through recalculation based on the reported measures of accuracy in combination with the prevalence in and sample size of the included study. We calculated the pooled sensitivity, specificity, positive likelihood ratio (PLR), and negative likelihood ratio (NLR) as pooled estimates with 95% confidence intervals (CIs). 10 The likelihood ratios represent the likelihood that an index test result would be expected in a patient with a certain disease as compared with the likelihood of that same result among patients without that disease. PLR was calculated by dividing the pooled sensitivity by 1 − specificity, and NLR was calculated by dividing 1 − sensitivity by specificity. In the presence of significant heterogeneity by visual inspection of the forest plots, meta-regression analysis was performed to identify a potential source of bias using the following as covariates: study design (single-center vs. multicenter study), number of patients (<150 vs. 

The literature search process is shown in Figure 1 . We initially identified 138 articles from PubMed, 164 articles from Embase, 224 articles from the Cochrane COVID-19 Study Register, and one additional article from hand-searching. After removing duplicate articles, we screened 369 potentially eligible articles. After reviewing the title and abstracts, 316 search records were removed; the remaining 52 articles were eligible for full-text review. Thirty-one articles were excluded for the reasons shown in Figure 1 . With quantitative synthesis, 17 studies were included in our final analysis. [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] Table 1 summarizes the features of the included studies. For the Xpert Xpress assay, we identified 11 studies comprising 1734 participants. 13, [16] [17] [18] [19] [22] [23] [24] [25] 27, 28 Four studies were designed as multicenter trials, 14, 16, 19, 20 and 13 studies were conducted only in the United States. 12, 14, 15, 17, 18, [20] [21] [22] [23] [24] [25] [26] 28 The number of patients in each trial ranged from 26 to 524. For the ID NOW assay, 10 studies involving 1778 subjects met the defined inclusion criteria. 12, 14, 15, 17, 20, 21, 23, 24, 26, 28 Four studies assessed samples using both the Xpert Xpress and the ID NOW assays. 14, 17, 23, 26 As the reference standard, more than two types of RT-PCR methods were used in seven studies. 12, [17] [18] [19] [20] 23, 26 Eleven studies used NP swab specimens only. 12, 13, 15, [18] [19] [20] [21] [24] [25] [26] 28 The QUADAS-2 assessment results are presented in Figures S1 and S2. The included studies were generally at high or unclear risk of bias. Patient selection procedures involved mostly high or unclear risk of bias, as most studies did not describe a consecutive cohort.

The index tests involved mostly unclear risk of bias, because it was frequently unclear whether the index tests were interpreted without knowledge of the results of the reference standard. For the reference standard domain, we mostly judged the risk of bias as SARS-CoV-2 infection, which is low enough to exclude SARS-CoV-2 in the clinical setting. 29 The pooled NLR was 0.21, which is a concern for exclusion of SARS-CoV-2. 29 A possible explanation for the low sensitivity of the ID NOW assay is the difference in detection accuracy related to viral burden. 24 A recent study compared the Xpert Xpress assay and the ID NOW assay with the Roche Cobas SARS-CoV-2 assay for samples with low, medium, and high SARS-CoV-2 viral concentrations. 22 to 52.2%) for the ID NOW assay. 24 In addition, a short article concerning the ID NOW assay reported that all false-negative results were recorded from samples that had low viral concentration with C t values between 35 and 40. 20 Another trial indicated that false-negative rates of the ID NOW assay exceeded 10% for samples with low viral concentrations. 23 On the basis of these findings, the ID NOW assay seems to have insufficient sensitivity for samples with low viral concentrations.

To the best of our knowledge, four molecular tests have currently been approved for diagnosis of the SARS-CoV-2 virus under EUA from the FDA. 5 We first tried to investigate the diagnostic accuracy of the approved rapid molecular tests. However, as the Accula SARS-CoV-2 rapid molecular test was used in only one study, we did not include this test in our analysis. 30 Meanwhile, the sensitivity and specificity of the Accula SARS-CoV-2 test were reported to be 0.68 (95% CI, 0.53 to 0.81) and 1.00 (95% CI, 0.93 to 1.00), respectively. 30 Potential limitations of the present study should be considered when interpreting our results. First, the studies included in our metaanalysis were of low methodological quality based on QUADAS-2 assessment. As the studies often lacked reporting of key information, assessment of several criteria for risk of bias was "high" or "unclear."

Therefore, our results should be carefully interpreted due to limited methodological quality. Second, we could not assess publication bias as no reliable methods exist to investigate this in diagnostic test accuracy studies. 32 Third, we used various types of commercial or laboratory-based RT-PCRs for reference comparisons in SARS-CoV-2 diagnosis, which can introduce bias due to diagnostic differences. Finally, significant heterogeneity was found in both the sensitivity and specificity of the ID NOW assay. Although metaregression analysis was performed, we could not identify potential sources of bias.

We demonstrated the excellent diagnostic accuracy of the Xpert Xpress assay as a robust diagnostic method for point-of-care diagnosis of SARS-CoV-2. On the contrary, as the ID NOW assay had relatively low pooled sensitivity, this test might miss several positive patient specimens. Therefore, some patients with negative results on the ID NOW assay require another confirmatory NAAT test.

Individuals requiring rapid confirmation of SARS-CoV-2 infection might benefit from the findings of this study.

Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan, China

SARS-CoV-2 is an appropriate name for the new coronavirus

Effect of delay in diagnosis on transmission of COVID-19

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

Food and Drug Administration

Preferred Reporting Items for a Systematic Review and Meta-analysis of diagnostic test accuracy studies: the PRISMA-DTA Statement

The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews

Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews

Empirical evidence of design-related bias in studies of diagnostic tests

Clinical evaluation and utilization of multiple molecular in vitro diagnostic assays for the detection of SARS-CoV-2

Brief validation of the novel GeneXpert Xpress SARS-CoV-2 PCR assay

Comparison of Abbott ID Now and Abbott m2000 methods for the detection of SARS-CoV-2 from nasopharyngeal and nasal swabs from symptomatic patients

Five-minute point-of-care testing for SARS-CoV-2: not there yet

Multicenter evaluation of the Cepheid Xpert Xpress SARS-CoV-2 assay for the detection of SARS-CoV-2 in oropharyngeal swab specimens

Comparative study of four SARS-CoV-2 nucleic acid amplification test (NAAT) platforms demonstrates that ID NOW performance is impaired substantially by patient and specimen type

Comparison of commercially available and laboratory-developed assays for in vitro detection of SARS-CoV-2 in clinical laboratories

Multicenter evaluation of the Cepheid Xpert Xpress SARS-CoV-2 test

Evaluation of the COVID19 ID NOW EUA assay

Comparison of two commercial molecular tests and a laboratory-developed modification of the CDC 2019-nCoV reverse transcriptase PCR assay for the detection of SARS-CoV-2

Detection of SARS-CoV-2 by use of the Cepheid Xpert Xpress SARS-CoV-2 and Roche cobas SARS-CoV-2 assays

A Comparison of five SARS-CoV-2 molecular assays with clinical correlations

Comparison of Cepheid Xpert Xpress and Abbott ID Now to Roche cobas for the rapid detection of SARS-CoV-2

Comparison of a point-ofcare assay and a high-complexity assay for detection of SARS-CoV-2 RNA

How many are we missing with ID NOW COVID-19 assay using direct nasopharyngeal swabs? Findings from a mid-sized academic hospital clinical microbiology laboratory

Evaluation on testing of deep throat saliva and lower respiratory tract specimens with Xpert Xpress SARS-CoV-2 assay

Clinical evaluation of three sample-to-answer platforms for detection of SARS-CoV-2

Simplifying likelihood ratios

Comparison of the Accula SARS-CoV-2 test with a laboratory-developed assay for detection of SARS-CoV-2 RNA in clinical nasopharyngeal specimens

Rapid, point-of-care antigen and molecular-based tests for diagnosis of SARS-CoV-2 infection

Systematic reviews and meta-analyses of diagnostic test accuracy

The authors would like to thank Sung Ryul Shim for assistance with statistical analysis. This work was supported by a research grant from the Jeju National University Hospital Research Fund of Jeju National University School of Medicine in 2020.

The authors declare that there are no conflict of interests.

Jae-Uk Song contributed to data acquisition, data interpretation, statistical analysis, and drafted the manuscript. Jonghoo Lee contributed to the study design, data acquisition, data interpretation, statistical analysis, writing of the manuscript, and critical revision of the manuscript.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

http://orcid.org/0000-0003-2626-7099Jae-Uk Song http://orcid.org/0000-0003-4597-7037