key: cord-1041711-qj4svnye
authors: Ridgway, Jessica P; Pisano, Jennifer; Landon, Emily; Beavis, Kathleen G; Robicsek, Ari
title: Clinical Sensitivity of Severe Acute Respiratory Syndrome Coronavirus 2 Nucleic Acid Amplification Tests for Diagnosing Coronavirus Disease 2019
date: 2020-07-24
journal: Open Forum Infect Dis
DOI: 10.1093/ofid/ofaa315
sha: 96e268f9a4275ac4b1cd89d9f2e50252f222be67
doc_id: 1041711
cord_uid: qj4svnye

Utilizing 34 348 severe acute respiratory syndrome coronavirus 2 (SARS CoV-2) nucleic acid amplification test (NAAT) results from 2 health systems, we estimated the clinical sensitivity of a single SARS-CoV-2 NAAT. We found that SARS-CoV-2 NAAT has 82%–97% sensitivity for diagnosing coronavirus disease 2019 among symptomatic patients.

In response to the coronavirus disease 2019 (COVID-19) pandemic, the US Food and Drug Administration issued emergency authorization for use of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) nucleic acid amplification tests (NAATs) to diagnose COVID-19 [1] . Despite widespread use of SARS-CoV-2 molecular testing, its clinical sensitivity remains uncertain. Reports of false-negative results among patients with COVID-19 underscore the need for systematic study of the test's sensitivity in a real-world setting [2, 3] . METHODS We used 2 methods to calculate the clinical sensitivity of a single SARS-CoV-2 NAAT.

We collected test results for all symptomatic patients tested with SARS-CoV-2 NAAT via nasopharyngeal swab at University of Chicago Medicine (UCM) and Providence St. Joseph Health between January 22, 2020 and April 23, 2020. During the study period, UCM only tested patients with symptoms consistent with COVID-19. Because Providence includes 51 hospitals with different policies around COVID-19 testing (including some that were screening asymptomatic patients), we only included Providence test data for patients with at least 2 of the following 3 symptoms: fever, cough, and shortness of breath. Symptom data were extracted from the electronic medical record using natural language processing of clinical notes.

Several different SARS-CoV-2 NAATs were used during the study period. Tests used at UCM were Cepheid Xpert Xpress SARS-CoV-2 and Roche cobas SARS-CoV-2. Tests used at Providence included Abbott ID Now COVID-19, BD SARS-CoV-2 Reagents for BD Max System, BioFire COVID-19, BioGX SARS-CoV-2 reagents for BD Max System, CDC 2019-nCoV RT-PCR Diagnostic Panel, Cepheid Xpert Xpress SARS-CoV-2, LabCorp COVID-19 RT-PCR, Panther Fusion SARS-CoV-2, Roche cobas SARS-CoV-2, Quest SARS-CoV-2 rRT-PCR, and Simplexa COVID-19 Direct.

To calculate sensitivity, we made several assumptions. We assumed all positive SARS-CoV-2 NAATs represented truepositive (TP) results. Among patients testing negative who were subsequently retested within 7 days of their first test, we assumed that an initial negative test followed by a positive test represented a false-negative (FN) result, and that an initial negative test followed by a second negative test represented a truenegative (TN) result. Because many patients were tested only once for SARS-CoV-2, we assumed that the FN rate ([FNR] calculated as FN/(FN + TN)) of a single negative NAAT was the same among patients tested once as it was among patients tested multiple times. Based on this assumption, the number of FN results in the entire population of tested patients was calculated as FN = number of initial negative tests × FNR. Sensitivity was calculated as TP/(TP + FN).

For the second sensitivity calculation, we limited our analysis to patients hospitalized at UCM during the study period. The UCM requires that inpatients with a negative test for SARS-CoV-2 undergo repeat testing at least 48 hours after their initial test. This 2-test policy is in place to ensure that patients with possible COVID-19 remain on appropriate infection control precautions in case of an FN result. We made the same assumptions regarding TP and FN as in Method 1. However, because Method 2 only included test results from a population of inpatients routinely tested twice for SARS-CoV-2, we were able to accurately identify FN test results without further assumptions. Sensitivity was calculated as TP/(TP + FN).

This study was approved by Institutional Review Boards (IRBs) at University of Chicago and Providence St. Joseph Health. Informed consent was not obtained because a waiver of consent was granted by the IRBs.

Over the study period, 34 348 SARS CoV-2 NAATs were performed ( Figure 1) . A total of 11.8% (4037 of 34 348) were positive. Of the 30 311 negative SARS-CoV-2 NAATs, 5.9% (1774 of 30 311) were followed by a subsequent SARS-CoV-2 NAAT within 7 days. Of these subsequent tests, 53 were positive, indicating a FN rate of 3.0% (53 of 1774). Estimated FN results in the entire tested population was calculated as 30 311*(53/1774) = 906. Sensitivity was calculated as 4037/ (4037 + 906) = 82% (Table 1) .

During the study period, 2443 SARS CoV-2 NAATs were performed among patients hospitalized at UCM. Four hundred thirty-seven tests were initially positive (TP). Fifteen negative tests were followed by a positive test within 7 days (FN). Sensitivity was calculated as 437/(437 + 15) = 97%.

In this multicenter US study, we found that SARS-CoV-2 NAAT has clinical sensitivity of 82%-97% for diagnosing COVID-19 among symptomatic patients. Although the test manufacturers have estimated clinical sensitivity using contrived clinical specimens, ours is the first large study to estimate SARS-CoV-2 NAAT clinical sensitivity in a real-world setting [4] .

In the absence of a gold standard for diagnosing COVID-19, we estimated sensitivity based on several assumptions. In our first sensitivity calculation, we assumed the same FN rate among patients tested once and patients tested multiple times for SARS-CoV-2. In reality, patients with a single negative test result may have a lower likelihood of COVID-19 than patients who undergo repeat testing. Patients tested multiple times may have persistent symptoms consistent with COVID-19, prompting their medical providers to repeat testing. Therefore, our sensitivity estimate of 82% is best understood as the lower limit of sensitivity. If we assume that patients undergoing repeat testing have twice the likelihood of COVID-19 as patients who are not retested (i.e, repeat testers have twice the FN rate as patients tested once), then the sensitivity estimate increases to 89% (4037 of 4516). If we assume that repeat testers have 3 times the likelihood of COVID-19, then the sensitivity estimate increases to 92% (4037 of 4374).

We also assumed that negative NAATs followed by positive tests within 7 days were FN results. In some cases, patients may have acquired COVID-19 within the 7 days between tests. However, when we limited our analysis to tests repeated within 3 days, our results were similar (data not shown). Finally, we assumed all positive test results were TP results. Although it is possible that some of the positive SARS-CoV-2 NAAT results were false positives, it is unlikely that this would alter our results given the high analytical specificity of the SARS-CoV-2 NAATs used.

Beyond the assumptions made, our study had some additional limitations. Eleven different NAATs were used during the study period. with reverse-transcription polymerase chain reaction tests [5, 6] . In addition, we were not able to collect information regarding timing of symptom onset related to testing. We limited our analysis to symptomatic patients, and the clinical sensitivity of SARS-CoV-2 NAAT for diagnosis of COVID-19 among asymptomatic patients may be different. Our analysis only included nasopharyngeal samples tested for SARS-CoV-2. For patients with symptoms of COVID-19 with an initial negative nasopharyngeal SARS CoV-2 NAAT result, obtaining a bronchoalveolar lavage for SARS-CoV-2 testing may have a higher diagnostic yield compared with repeating another nasopharyngeal testx.

In conclusion, this study estimated clinical sensitivity of NAAT for clinical diagnosis of COVID-19 among symptomatic patients in a real-world setting among over 30 000 patients. Our findings of a relatively high sensitivity of a single SARS-CoV-2 NAAT have important implications for clinical diagnosis of COVID-19.

Policy for Diagnostics Testing in Laboratories Certified to Perform High-Complexity Testing under Clinical Laboratory Improvement Amendments (CLIA) prior to Emergency Use Authorization for Coronavirus Disease-2019 during the

Diagnosis of the coronavirus disease (COVID-19): rRT-PCR or CT?

Sensitivity of chest CT for COVID-19: comparison to RT-PCR

Sensitivity" and "specificity" reconsidered: the meaning of these terms in analytical and diagnostic settings

Performance of Abbott ID NOW COVID-19 rapid nucleic acid amplification test in nasopharyngeal swabs transported in viral media and dry nasal swabs, in a New York City academic institution

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

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.