Original Research18 August 2020

Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests by Time Since Exposure

FREE
    Author, Article and Disclosure Information

    Abstract

    Background:

    Tests for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) based on reverse transcriptase polymerase chain reaction (RT-PCR) are being used to rule out infection among high-risk persons, such as exposed inpatients and health care workers. It is critical to understand how the predictive value of the test varies with time from exposure and symptom onset to avoid being falsely reassured by negative test results.

    Objective:

    To estimate the false-negative rate by day since infection.

    Design:

    Literature review and pooled analysis.

    Setting:

    7 previously published studies providing data on RT-PCR performance by time since symptom onset or SARS-CoV-2 exposure using samples from the upper respiratory tract (n = 1330).

    Patients:

    A mix of inpatients and outpatients with SARS-CoV-2 infection.

    Measurements:

    A Bayesian hierarchical model was fitted to estimate the false-negative rate by day since exposure and symptom onset.

    Results:

    Over the 4 days of infection before the typical time of symptom onset (day 5), the probability of a false-negative result in an infected person decreases from 100% (95% CI, 100% to 100%) on day 1 to 67% (CI, 27% to 94%) on day 4. On the day of symptom onset, the median false-negative rate was 38% (CI, 18% to 65%). This decreased to 20% (CI, 12% to 30%) on day 8 (3 days after symptom onset) then began to increase again, from 21% (CI, 13% to 31%) on day 9 to 66% (CI, 54% to 77%) on day 21.

    Limitation:

    Imprecise estimates due to heterogeneity in the design of studies on which results were based.

    Conclusion:

    Care must be taken in interpreting RT-PCR tests for SARS-CoV-2 infection—particularly early in the course of infection—when using these results as a basis for removing precautions intended to prevent onward transmission. If clinical suspicion is high, infection should not be ruled out on the basis of RT-PCR alone, and the clinical and epidemiologic situation should be carefully considered.

    Primary Funding Source:

    National Institute of Allergy and Infectious Diseases, Johns Hopkins Health System, and U.S. Centers for Disease Control and Prevention.

    Tests for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) based on reverse transcriptase polymerase chain reaction (RT-PCR) are often used to rule out infection among high-risk persons, such as exposed inpatients and health care workers. Hence, it is critical to understand how the predictive value changes in relation to time since exposure or symptoms, especially when using the results of these tests to make decisions about whether to stop using personal protective equipment or allow exposed health care workers to return to work. The sensitivity and specificity of PCR-based tests for SARS-CoV-2 are poorly characterized, and the “window period” after acquisition in which testing is most likely to produce false-negative results is not well known.

    Accurate testing for SARS-CoV-2, followed by appropriate preventive measures, is paramount in the health care setting to prevent both nosocomial and community transmission. However, most hospitals are facing critical shortages of SARS-CoV-2 testing capacity, personal protective equipment, and health care personnel (1). As the epidemic progresses, hospitals increasingly have to decide how to respond when a patient or health care worker has a known exposure to SARS-CoV-2. Although 14 days of airborne precautions or quarantine would be a conservative approach to minimizing transmission per guidelines from the Centers for Disease Control and Prevention (2), this is not feasible for many hospitals given starkly limited resources.

    As RT-PCR–based tests for SARS-CoV-2 are becoming more available, they are increasingly being used to rule out infection to conserve scarce personal protective equipment and preserve the workforce. When exposed health care workers test negative, they may be cleared to return to work; similarly, when exposed patients test negative, airborne or droplet precautions may be removed. If negative results from tests done during the window period are treated as strong evidence that an exposed person is SARS-CoV-2–negative, preventable transmission could occur.

    It is critical to understand how the predictive value of the test varies with time from exposure and symptom onset to avoid being falsely reassured by negative results from tests done early in the course of infection. The goal of our study was to estimate the false-negative rate by day since infection.

    Methods
    Source Data

    As part of a broader effort to provide critical evaluation of emerging evidence, the Novel Coronavirus Research Compendium at the Johns Hopkins School of Public Health did a literature review to identify preprint and peer-reviewed articles on SARS-CoV-2 diagnostics (3). Investigators searched PubMed, bioRxiv, and medRxiv using a strategy detailed in Supplement Table 1. The search was last updated on 15 April 2020. From the broader search, we identified articles that provided data on RT-PCR performance by time since symptom onset or exposure using samples derived from nasal or throat swabs among patients tested for SARS-CoV-2. Inclusion criteria were use of an RT-PCR–based test, sample collection from the upper respiratory tract, and reporting of time since symptom onset or exposure. We excluded articles that did not clearly define time between testing and symptom onset or exposure. We identified 7 studies (2 preprints and 5 peer-reviewed articles) (4–10) with a total of 1330 respiratory samples analyzed by RT-PCR. Figure 1 summarizes the source data. One study by Kujawski and colleagues (10) provided both nasal and throat samples for each patient; we used only the nasal samples in our analysis.

    Figure 1. Sensitivity of RT-PCR tests, by study and days since symptom onset, for nasopharyngeal samples (left), oropharyngeal samples (middle), and unspecified upper respiratory tract (right).

    RT-PCR = reverse transcriptase polymerase chain reaction.

    How Cases Were Defined

    Most studies (Danis and colleagues [6], Wölfel and colleagues [4], Kim and colleagues [7], Kujawski and colleagues [10], and Zhao and colleagues [8]) did serial testing and required at least 1 positive RT-PCR result to consider a case confirmed. Our pooled analysis included only confirmed cases from those studies. The studies by Liu and colleagues (9) and Guo and colleagues (5) included both confirmed cases (≥1 positive RT-PCR result, similar to other studies; n = 153 for Liu and n = 82 for Guo) and probable cases as determined by a set of clinical criteria (n = 85 for Liu and n = 58 for Guo). In both studies, most probable case patients were positive for IgM or IgG SARS-CoV-2 antibodies (67 of 85 probable cases for Liu were IgM- or IgG-positive, and 54 of 58 for Guo were IgM-positive). Thus, 22 participants were considered case patients on the basis of clinical criteria alone because we could not separate them out using the information provided. Supplement Table 2 provides additional details on the source data used in our calculations. As a sensitivity analysis to assess the effect of individual studies on our inferences, we excluded each study in turn from calculations of the posttest probability of infection after a negative RT-PCR result (Supplement Figure 3).

    Statistical Analysis
    Model for Estimating False-Negative Rate and False Omission Rate by Time Since Exposure

    Using an approach similar to that of Leisenring and colleagues (11) and Azman and colleagues (12), we fitted a Bayesian hierarchical logistic regression model for test sensitivity pj,t with a random effect for study j and a cubic polynomial spline for log-time t since exposure:

    where xj,t is the number of patients who tested positive on RT-PCR out of nj,t total tests t days after exposure in study j. The exposure was assumed to have occurred 5 days before symptom onset based on the median incubation period previously estimated in a large study of transmission in household contacts (13) and among publicly confirmed cases (14). From the sensitivity, we calculated the expected false-negative rate on each day. We also calculated the posttest probability of infection, assuming a pretest probability based on the attack rate in close household contacts of SARS-CoV-2 case patients in Shenzhen, China (77 of 686 [11.2%]) (14). We assumed a specificity of 100% for RT-PCR, as reported in the U.S. Food and Drug Administration package insert for the Quest RT-PCR assay for SARS-CoV-2, which based its estimate on testing in 72 presumed negative samples from the upper respiratory tract and 30 from the lower respiratory tract (15). This specificity is further supported by a European study that showed no cross-reactivity with other coronaviruses in 297 clinical samples (16).

    Sensitivity Analyses

    Although the Food and Drug Administration reported that specificity for SARS-CoV-2 RT-PCR is 100%, many of the supporting studies were done outside the United States, and we cannot exclude variability in test performance. Thus, we repeated our analysis assuming 90% specificity to assess the sensitivity of our results to this assumption. A second assumption of our model, the 5-day incubation period, was based on a large study of household contacts in Shenzhen (13) and on publicly confirmed cases (14). We did additional analyses varying the incubation period to 3 and 7 days to assess the sensitivity of our results to this assumption. We also repeated analyses excluding 1 study each time to assess the effect on our inferences.

    Code and Data Availability

    The data and code used to run this analysis are publicly available at https://github.com/HopkinsIDD/covidRTPCR (17).

    Role of the Funding Source

    The funders had no influence on the study's design, conduct, or reporting.

    Results
    Probability of a False-Negative Result Among SARS-CoV-2–Positive Patients, by Day Since Exposure

    Over the 4 days of infection before the typical time of symptom onset (day 5), the probability of a false-negative result in an infected person decreases from 100% (95% CI, 100% to 100%) on day 1 to 67% (CI, 27% to 94%) on day 4, although there is considerable uncertainty in these numbers. On the day of symptom onset, the median false-negative rate was 38% (CI, 18% to 65%) (Figure 2, top). This decreased to 20% (CI, 12% to 30%) on day 8 (3 days after symptom onset) then began to increase again, from 21% (CI, 13% to 31%) on day 9 to 66% (CI, 54% to 77%) on day 21.

    Figure 2. Probability of having a negative RT-PCR test result given SARS-CoV-2 infection (top) and of being infected with SARS-CoV-2 after a negative RT-PCR test result (bottom), by days since exposure.

    RT-PCR = reverse transcriptase polymerase chain reaction; SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2.

    Posttest Probability of Infection if RT-PCR Result is Negative (1 Minus Negative Predictive Value)

    Translating these results into a posttest probability of infection, a negative result on day 3 would reduce our estimate of the relative probability that a case patient was infected by only 3% (CI, 0% to 47%) (for example, from 11.2%, the rate seen in a large study of household contacts, to 10.9%) (Figure 2, bottom). Tests done on the first day of symptom onset are more informative, reducing the inferred probability that a case patient was infected by 60% (CI, 33% to 80%).

    Variation in Posttest Probability of Infection if RT-PCR Result is Negative, by Pretest Probability

    The posttest probability of infection in a patient with a negative RT-PCR result varies with the pretest probability of infection—that is, how likely infection is on the basis of the magnitude of exposure or clinical presentation. When we assumed a high pretest probability of infection (4 times the attack rate observed in a large cohort study), the posttest probability of infection was at minimum 14% (CI, 9% to 20%) 8 days after exposure (Figure 3). When we assumed a lower pretest probability of 5.5% (half the observed attack rate), the negative posttest probability of infection was still minimized 8 days after exposure (1.2% [CI, 0.7% to 2.0%]).

    Figure 3. Posttest probability of SARS-CoV-2 infection after a negative RT-PCR result, by pretest probability of infection.

    RT-PCR = reverse transcriptase polymerase chain reaction; SARS-CoV-2 = severe acute respiratory syndrome coronavirus 2.

    Sensitivity Analyses

    When we repeated our analysis assuming a specificity of RT-PCR of 90% rather than 100%, results were very similar (Supplement Figure 1). We found a higher probability of infection in the setting of a negative RT-PCR result, with the greatest difference occurring on day 2 (12.4% vs. 11.3% [1.1 percentage point higher]). When we repeated our analyses varying the incubation period, we found that an earlier onset time of symptoms led to a quicker decrease in false omission rate and a later onset time led to a slower decrease; however, curves were similar overall, and our primary inferences remained the same relative to the date of onset (Supplement Figure 2). When we repeated our analysis of the posttest probability of infection excluding a different study each time, our inferences were unchanged (Supplement Figure 3).

    Discussion

    Over the 4 days of infection before the typical time of symptom onset (day 5), the probability of a false-negative result in an infected person decreased from 100% on day 1 to 68% on day 4. On the day of symptom onset, the median false-negative rate was 38%. This decreased to 20% on day 8 (3 days after symptom onset) then began to increase again, from 21% on day 9 to 66% on day 21. The false-negative rate was minimized 8 days after exposure—that is, 3 days after the onset of symptoms on average. As such, this may be the optimal time for testing if the goal is to minimize false-negative results. When the pretest probability of infection is high, the posttest probability remains high even with a negative result. Furthermore, if testing is done immediately after exposure, the pretest probability is equal to the negative posttest probability, meaning that the test provides no additional information about the likelihood of infection.

    Since the outbreak began, concerns have been raised about the poor sensitivity of RT-PCR–based tests (18); 1 study has suggested that this might be as low as 59% (19). We have designed a publicly available model that provides a framework for estimating the performance of these tests by time since exposure and can be updated as additional data become available.

    Tests for SARS-CoV-2 based on RT-PCR added little diagnostic value in the days immediately after exposure. This is consistent with a window period between acquisition of infection and detectability by RT-PCR seen in other viral infections, such as HIV and hepatitis C (20, 21). Our study suggests a window period of 3 to 5 days, and we would not recommend making decisions regarding removing contact precautions or ending quarantine on the basis of results obtained in this period in the absence of symptoms. Although the false-negative rate is minimized 1 week after exposure, it remains high at 21%. Possible mechanisms for the high false-negative rate include variability in individual amount of viral shedding and sample collection techniques.

    One consideration is whether serial testing would offer any benefit in test performance compared with a single test. If we assume independence of the test results, serial testing would almost certainly reduce the false-negative rate; however, without more data on the underlying mechanism for the high false-negative rate, this assumption may not be warranted. For example, if the rate were due to individual variability in viral shedding, performance would likely not be improved by serial tests. Although we are aware of no large-scale studies, some preliminary reports suggest lack of independence; for example, in 1 case report of a person with infection confirmed on the basis of both radiologic findings and RT-PCR positivity from endotracheal aspirates, RT-PCR results from nasopharyngeal swabs were negative throughout the clinical course (6). Further studies to better characterize the underlying mechanism for poor diagnostic performance of SARS-CoV-2 RT-PCR are needed to inform testing strategies.

    The relationship between a false-negative result and infectiousness is unclear, and patients who test negative on samples from nasopharyngeal swabs may be less likely to transmit the virus regardless of true case status. We found an increase in the false-negative rate starting 9 days after exposure; however, it is possible that some of the later results were not actual false negatives but rather represented clearance of the infection. Thus, interpretation later in the clinical course depends on the purpose of testing: If the goal is to clear a patient from isolation, these negative results may be correct, although more data are needed given studies showing viral replication in other sites. However, if the goal of the test is to evaluate whether additional follow-up is needed or whether the patient should be treated as SARS-CoV-2–positive for the purpose of contact tracing, the test may not be providing the desired information and caution should be used in decision making. Because antibodies appear later in the course of infection, a combination of antibody testing and RT-PCR might be most useful for patients more remote from symptoms or exposure.

    Our study has several limitations. There was significant heterogeneity in the design and conduct of the underlying studies from which the data used in our analyses were drawn. However, when we did a sensitivity analysis excluding each study in turn, we found that no 1 study was especially influential and inferences were largely unchanged. Sample collection techniques varied across studies (oropharyngeal vs. nasopharyngeal swabs), and several studies stated that samples were from the upper respiratory tract without providing further details. Thus, we could not fully account for differences in sample collection techniques. Most studies tested samples at time of symptom onset rather than time of exposure, leading to high variance in estimates in the first few days after exposure. Our model is applicable only in the setting of a known, one-time exposure, not in the setting of continuous exposure, such as in health care workers who may be exposed daily to SARS-CoV-2–positive patients. Finally, most studies defined true-positive cases as those with at least 1 positive RT-PCR result, meaning that patients who never tested positive would not be included; this could lead to underestimation of the true false-negative rate. Two studies included probable cases based on clinical and epidemiologic characteristics even if the patients had never had a positive RT-PCR result or serology. Because such criteria as fever, respiratory symptoms, and imaging findings are nonspecific, misclassification is likely, wherein some proportion of probable cases are actually true negatives rather than false negatives. We believe that this effect was small because excluding these studies from our analysis did not change our primary inferences.

    In summary, care must be taken when interpreting RT-PCR tests for SARS-CoV-2 infection, particularly early in the course of infection and especially when using these results as a basis for removing precautions intended to prevent onward transmission. If clinical suspicion is high, infection should not be ruled out on the basis of RT-PCR alone, and the clinical and epidemiologic situation should be carefully considered. In many cases, time of exposure is unknown and testing is done on the basis of time of symptom onset. The false-negative rate is lowest 3 days after onset of symptoms, or approximately 8 days after exposure. Clinicians should consider waiting 1 to 3 days after symptom onset to minimize the probability of a false-negative result. Further studies to characterize test performance and research into higher-sensitivity approaches are critical.

    References

    • 1. The LancetCOVID-19: protecting health-care workers [Editorial]. Lancet2020;395:922. [PMID: 32199474] doi:10.1016/S0140-6736(20)30644-9 CrossrefMedlineGoogle Scholar
    • 2. Centers for Disease Control and Prevention. Coronavirus disease 2019 (COVID-19). 2020. Accessed at www.cdc.gov/coronavirus/2019-ncov/hcp/guidance-risk-assesment-hcp.html on 31 March 2020. Google Scholar
    • 3. Johns Hopkins Bloomberg School of Public Health. 2019 Novel Coronavirus Research Compendium (NCRC). 2020. Accessed at https://ncrc.jhsph.edu on 8 May 2020. Google Scholar
    • 4. Wölfel R Corman VM Guggemos W et alVirological assessment of hospitalized patients with COVID-2019. Nature2020. [PMID: 32235945] doi:10.1038/s41586-020-2196-x CrossrefMedlineGoogle Scholar
    • 5. Guo L Ren L Yang S et alProfiling early humoral response to diagnose novel coronavirus disease (COVID-19). Clin Infect Dis2020. [PMID: 32198501] doi:10.1093/cid/ciaa310 CrossrefMedlineGoogle Scholar
    • 6. Danis K Epaulard O Bénet T et alInvestigation TeamCluster of coronavirus disease 2019 (Covid-19) in the French Alps, 2020. Clin Infect Dis2020. [PMID: 32277759] doi:10.1093/cid/ciaa424 CrossrefMedlineGoogle Scholar
    • 7. Kim ES Chin BS Kang CK et alKorea National Committee for Clinical Management of COVID-19Clinical course and outcomes of patients with severe acute respiratory syndrome coronavirus 2 infection: a preliminary report of the first 28 patients from the Korean cohort study on COVID-19. J Korean Med Sci2020;35:e142. [PMID: 32242348] doi:10.3346/jkms.2020.35.e142 CrossrefMedlineGoogle Scholar
    • 8. Zhao J Yuan Q Wang H et alAntibody responses to SARS-CoV-2 in patients of novel coronavirus disease 2019. Clin Infect Dis2020. [PMID: 32221519] doi:10.1093/cid/ciaa344 CrossrefMedlineGoogle Scholar
    • 9. Liu L, Liu W, Wang S, et al. A preliminary study on serological assay for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in 238 admitted hospital patients. Preprint. Posted online 8 March 2020. medRxiv. doi:10.1101/2020.03.06.20031856 Google Scholar
    • 10. Kujawski SA, Wong KK, Collins JP, et al. First 12 patients with coronavirus disease 2019 (COVID-19) in the United States. Preprint. Posted online 12 March 2020. medRxiv. doi:10.1101/2020.03.09.20032896 Google Scholar
    • 11. Leisenring W Pepe MS Longton G A marginal regression modelling framework for evaluating medical diagnostic tests. Stat Med1997;16:1263-81. [PMID: 9194271] CrossrefMedlineGoogle Scholar
    • 12. Azman AS, Lauer S, Bhuiyan MTR, et al. Vibrio cholerae O1 transmission in Bangladesh: insights from a nationally-representative serosurvey. Preprint. Posted online 16 March 2020. medRxiv. doi:10.1101/2020.03.13.20035352 Google Scholar
    • 13. Lauer SA Grantz KH Bi Q et alThe incubation period of coronavirus disease 2019 (COVID-19) from publicly reported confirmed cases: estimation and application. Ann Intern Med2020. [PMID: 32150748]. doi:10.7326/M20-0504 LinkGoogle Scholar
    • 14. Bi Q, Wu Y, Mei S, et al. Epidemiology and transmission of COVID-19 in Shenzhen China: analysis of 391 cases and 1,286 of their close contacts. Preprint. Posted online 27 March 2020. medRxiv. doi:10.1101/2020.03.03.20028423 Google Scholar
    • 15. Quest Diagnostics. SARS-CoV-2 RNA, qualitative real-time RT-PCR (test code 39433): package insert. Accessed at www.fda.gov/media/136231/download on 20 April 2020. Google Scholar
    • 16. Corman VM Landt O Kaiser M et alDetection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR. Euro Surveill2020;25. [PMID: 31992387] doi:10.2807/1560-7917.ES.2020.25.3.2000045 CrossrefMedlineGoogle Scholar
    • 17. Kucirka LM, Lauer SA, Laeyendecker O, et al. Analysis of RT-PCR sensitivity by day since exposure or symptom onset. 2020. Accessed at https://github.com/HopkinsIDD/covidRTPCR on 26 April 2020. Google Scholar
    • 18. Krumholz HM. If you have coronavirus symptoms, assume you have the illness, even if you test negative. The New York Times. 1 April 2020. Accessed at www.nytimes.com/2020/04/01/well/live/coronavirus-symptoms-tests-false-negative.html on 20 April 2020. Google Scholar
    • 19. Ai T Yang Z Hou H et alCorrelation of chest CT and RT-PCR testing in coronavirus disease 2019 (COVID-19) in China: a report of 1014 cases. Radiology2020:200642. [PMID: 32101510] doi:10.1148/radiol.2020200642 CrossrefMedlineGoogle Scholar
    • 20. Konrad BP Taylor D Conway JM et alOn the duration of the period between exposure to HIV and detectable infection. Epidemics2017;20:73-83. [PMID: 28365331] doi:10.1016/j.epidem.2017.03.002 CrossrefMedlineGoogle Scholar
    • 21. Glynn SA Wright DJ Kleinman SH et alDynamics of viremia in early hepatitis C virus infection. Transfusion2005;45:994-1002. [PMID: 15934999] CrossrefMedlineGoogle Scholar

    Comments

    Gaurav Chhaya21 September 2021
    Nicely shown stastitical data on CoVID.

    Thanks for posting such elaborate statistics on SARS-COV-2. We have noticed such results in last year or so. And have gone with HRCT reports for same.  
    It is indeed needed to know such values and must be clinically judged on diseased state. 

    Gaurav Chhaya17 June 2021
    Good review … on COVID-19 false negative report

    Thanks for such an elaborate review of a subject which has many areas of unseen universe. It also was helpful in knowing this topic at large. 
    We at Indian areas go for HRCT Thorax if we were in doubt of result. 
    Many time it is seen that inadequate sample collection or collected with improper techniques create such reports dilemma. 
    Thank you again. 

    Lauren Kucirka16 June 2020
    Authors' Response to Comment Posted by Takhar et al

    In response to online comments raised by Takhar et al. (May 29, 2020), We understand that there may have been studies that we missed, although the paper by Young et al was published after our search period of Arpil 15, 2020. This is why, as the commenter’s state, we made all code and data publicly available so that the analysis can be updated as more studies are published. We have re-run the analysis with the data from Young et al. and have found no qualitative differences in results.

    We recognize that there are legitimate potential criticisms of extrapolating from trends occurring after symptom-onset to make estimates about the pre-symptomatic period; however, we felt that these extrapolations carried a very important clinical message in terms of the likely high false negative rate and high level of uncertainty about the performance of the test in this period. As more studies begin to report data on testing in pre-symptomatic individuals, it will be important to update these results.

    We recognize that, in retrospect, there are some modeling assumptions we could have made to further highlight the uncertainty of estimates in this period; for instance, treating the time of symptom onset as a random variable. These changes would have increased represented uncertainty intervals in certain periods, but not changed the core results.

    We respect the commenters’ difference in opinion about the utility of this modeling but continue to believe it provides important insights that we hope will be further informed by later work.

    Yutaka Hamaoka10 June 2020
    Methodological Limitations

    Comment: I read with great interests Kucirka et al.(1) that estimated the false-negative rate by day since infection based on seven published studies and estimated the probability of a false-negative result in an infected person on day 1 is 100% (95% CI, 100% to 100%). I identified two major limitations in the paper that could reject their conclusion.

    Firstly, although, they estimated the probability of a false-negative before onset, their data include only one before-onset case that reported in Danis et al.(2). Estimation based on a small sample could be unreliable and robustness of the results should be examined. In the Sensitivity analysis section, they reported unchanged false-positive curves for repeated analysis excluding one out of seven studies in the Supplement Figure 3. However, the credible interval of estimates that bring important information on robustness of the results is not reported and curves for false-negative is not displayed. I replicated analysis using their code and data (3), then obtained the probability of a false-negative result in an infected person on day 1 is 100% with widened 95% CI of (39.6% to 100%) excluding Danis et al.. Moreover, I examined the estimates of their model that determines the shape of daily false-negative and false-positive curves. As showed below, sign of coefficient revers for log(t), magnitude of coefficient are drastically changed for log(t)^2 and log(t)^3, and statistical significance of coefficient changed for log (t) and log(t)^2.   All data logit(pj,t) = -0.12 4.25 log(t) - 31.76 log(t)^2 14.12 log(t)^3 (95%CI= -0.63, 0.40) (-6.32, 15.48) (-48.72, -15.54) (4.82, 23.93) Excluding Danis et al logit(pj,t) = 0.13 - 11.69 log(t) -2.03 log(t)^2 2.57 log (t)^3 (-0.46, 0.70) (-14.30, -9.13) (-4.35, 0.29) (0.17, 4.91)

    Secondly, the paper neglects the nature of data that leads to model misspecification and biased estimates. According to Supplement Table 2 that summarize source data, original papers reported results of daily testing or repeated measure of subjects. In spite of that, results are tabulated into study by days from onset, then hierarchical regression model that incorporate study heterogeneity is estimated that assumes test results from the same subject are independent. Enough size of pre-onset cases, proper data handling and modeling that reflect nature of data are necessary to obtain reliable results. Careful model diagnosis must be conducted to reach a proper conclusion.

     

    Acknowledgement I appreciate Dr. Lauer SA. for publish data and script.

     

     Reference

    1. Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction-Based SARS-CoV-2 Tests by Time Since Exposure. Ann Intern Med. 2020.
    2. Danis K, Epaulard O, Benet T, Gaymard A, Campoy S, Bothelo-Nevers E, et al. Cluster of coronavirus disease 2019 (Covid-19) in the French Alps, 2020. Clin Infect Dis. 2020.
    3. Lauer SA. covidRTPCR. 2020: https://github.com/HopkinsIDD/covidRTPCR.
    Stijntje W. Dijk 1, Lisa Caulley 1 2, Emma K.T. Benn 3, Umut Özbek 3, Joost van Rosmalen 4, Bart S. Ferket 33 June 2020
    What is the Value of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests for Screening Asymptomatic Individuals?

    Kucirka and colleagues recently published an important analysis that accounts for time from exposure and symptom onset to obtain insights into the diagnostic performance of RT-PCR testing for infection with SARS-CoV-2 (1). Their analysis contributes to a better understanding of the sensitivity (1- false negative rate) of the RT-PCR test in the pre-symptomatic phase. The authors assume that test sensitivity is low (~0%) shortly after exposure and only increases to a reasonable value of ~60% at symptom onset (5 days later). The post-test probability of infection after a negative test remains substantial among pre-symptomatic individuals, especially when the population prevalence is high.

     There is growing support among governments, public health agencies, and healthcare systems worldwide to adopt universal screening strategies despite direct evidence of benefit from screening asymptomatic individuals. Guidelines for universal screening from the Infectious Disease Society of America (IDSA) are based on the assumption that the sensitivity of the RT-PCR test found in symptomatic individuals can be applied to asymptomatic individuals (2). The authors’ findings however cast doubt on the validity of IDSA’s assumption and raise questions that are critical to address before developing and implementing an effective strategy for screening asymptomatic individuals.

     First, the test sensitivity data from Kucirka and colleagues’ analysis only apply to the subgroup of pre-symptomatic individuals, i.e. those who develop symptoms later on, while widespread screening policies must be applicable to all asymptomatic individuals including up to 80% of infected individuals who never develop symptoms (3). Second, due to lack of data, the modeled sensitivity of the RT-PCR test in the pre-symptomatic phase depended on the assumed length of the incubation period. When assuming a 5-day period, the test sensitivity gradually declined to zero toward time of exposure (top panel in Figure 2). However, it is also plausible that shortly after exposure, sensitivity would increase more steeply and then level off (4). Lastly, the wide credible intervals limit the ability to assess diagnostic accuracy with confidence.

     To conclude, there remains considerable uncertainty around the sensitivity of the RT-PCR test in asymptomatic individuals. We concur with IDSA’s recommendations that as cities, countries and healthcare systems move towards reopening, well-designed diagnostic accuracy studies should become a high-priority research focus to better understand the variation in test sensitivity over the course of infection (5). The value of acquiring additional information from such studies could be used to inform decision models to more effectively guide screening policies.

     

     

    References

    1. Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests by Time Since Exposure. Ann Intern Med [Internet]. 2020 May 13; Available from: https://www.acpjournals.org/doi/abs/10.7326/M20-1495
    2. Hanson K, Caliendo M, Arias CA, Englund JA, Lee ML, Loeb M, et al. Infectious Diseases Society of America Guidelines on the Diagnosis of COVID-19 [Internet]. Arlington; Available from: https://www.idsociety.org/globalassets/idsa/practice-guidelines/covid-19/diagnostics/idsa-covid-19-guideline_dx_version-1.0.1.pdf
    3. World Health Organization. Q&A: Influenza and COVID-19 - similarities and differences [Internet]. [cited 2020 May 30]. Available from: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/question-and-answers-hub/q-a-detail/q-a-similarities-and-differences-covid-19-and-influenza
    4. He X, Lau EHY, Wu P, Deng X, Wang J, Hao X, et al. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nat Med. 2020;1–4.
    5. Cohen JF, Korevaar DA, Altman DG, Bruns DE, Gatsonis CA, Hooft L, et al. STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration. BMJ Open. 2016;6(11):e012799.

     

     

    sukhjit S. Takhar MD SM, David L. Schriger, MD MPH29 May 2020
    Refrain from modeling

    We agree with the spirit of article regarding a cautious approach in removing isolation precautions and using clinical suspicion along with RT-PCR for suspected SARS-CoV-2 infection. Kucirka et al(1) conclusion that deeming a patient COVID-free on the basis of a single negative RT-PCR test is fraught with peril, especially in those early in the course of the infection. We also appreciate that these authors posted their data and code and by so doing gave readers like us the opportunity to offer specific critique; this is the way science advances.

    We have two concerns. First, the authors did not include Young et al (2) which offered serial RT-PCR data on 18 patients in Singapore, leading us to wonder what other sources of data were missed. Second, and perhaps more important, we are concerned that data are so sparse that the results are based more on model assumptions than data. The authors identified 7 studies providing results of serial RT-PCR and time of symptom onset. They assume that exposure occurred 5 days prior to onset of symptoms. The authors then fit a bayesian hierarchical logistic model with a smoothing function in order to draw inference on the false-negative rate by day since exposure. Only one study, Danis et al (3) , had RT-PCR results for specimens obtained before the patients were symptomatic, however careful examination of that paper reveals that this consists of 4 sequential negative tests in a single patient (case 13) on days 2 to 5 before symptom onset. That is their entire data for the presymptomatic period.

    Given this, it is impossible to make any inferences for the presymptomatic days and it would have been wiser to refrain from modeling and simply say that these data cannot tell us when any positive results can be expected but that it is clear that there will be negative RT-PCR tests in both in the early symptomatic days and the pre-symptomatic days.

    1. Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction–Based SARS-CoV-2 Tests by Time Since Exposure. Annals of Internal Medicine. 2020 May 13.

    2. Young BE, Ong SW, Kalimuddin S, Low JG, Tan SY, Loh J, Ng OT, Marimuthu K, Ang LW, Mak TM, Lau SK. Epidemiologic features and clinical course of patients infected with SARS-CoV-2 in Singapore. Jama. 2020 Apr 21;323(15):1488-94.

    3. Danis K, Epaulard O, Bénet T, Gaymard A, Campoy S, Bothelo-Nevers E, Bouscambert-Duchamp M, Spaccaferri G, Ader F, Mailles A, Boudalaa Z. Cluster of coronavirus disease 2019 (Covid-19) in the French Alps, 2020. Clinical Infectious Diseases. 2020 Apr 11.

    Rudy Zimmer26 May 2020
    Delays in testing as a source of COVID-19 false negative results.

    The article by Kucirka and colleagues [1] highlights a major weakness in our reliance for diagnosis and public health surveillance on reverse transcriptase polymerase chain reaction (RT-PCR) based tests using nasopharyngeal (NP) or oropharyngeal (OP) sampling to determine community spread of “severe acute respiratory syndrome 2” (SARS-CoV-2) virus infection, which leads to “coronavirus disease 2019” (COVID-19). For mild to moderately ill patients, the window of reliable detection seems to be 7-10 days from onset of symptoms based on studies referenced by Kucirka [1], in addition to findings from Pan [2] and He [3].

    Government-run COVID-19 assessment centers (at least in Canada) have been fraught with booking delays from the time that symptomatic patients call 811 for an appointment to the time of actual testing. In my experience, patients have reported waiting 5 days on average (ranging from 3 days to 2 weeks), while already feeling unwell for several days. Systemic delays in testing among probable community cases raises the risk of preventable false negative results. The following example illustrates this problem in practice.

    On 22 April 2020, three patients called our office requesting COVID-19 testing, since they were still waiting to receive a centralized booking. All three were brought into the clinic immediately and tested using NP viral swab kits left over from the annual influenza surveillance program. Two of three patients were reported positive for SARS-CoV-2 infection the following day. One of the confirmed cases was a parent with a spouse and two children at home with similar symptoms. The older child started to become sick at the same time as the case, while the spouse became ill a few days later followed by the younger child. As the clinic did not have sufficient kits for testing the rest of the family, they contacted the local assessment center. When tested on 1 May 2020 (7 days later), the older child was “negative”, the spouse was initially “indeterminate” (later classified as “positive”) and the younger child was “positive”. Detection of SARS-CoV-2 virus in the upper airway appears to be short-lived in most moderately ill cases similar to that of influenza viruses [3].

    Decentralizing testing back into community-based clinics as part of routine medical care and reducing backlogs in centralized assessment centers by not testing persons with symptoms beyond 7-10 days would reduce the risk of false negative tests. As we open up the economy, we require smarter surveillance that better tracks community spread.

    [1] Kucirka LM, Lauer SA, Laeyendecker O, Boon D, Lessler J. Variation in False-Negative Rate of Reverse Transcriptase Polymerase Chain Reaction-Based SARS-CoV-2 Tests by Time Since Exposure. Ann Intern Med. 2020 May 13. doi: 10.7326/M20-1495. Available: https://www.acpjournals.org/doi/full/10.7326/M20-1495 (accessed 2020 May 25).

    [2] Pan Y, Zhang D, Yang P, Poon LLM, Wang Q. Viral load of SARS-CoV-2 in clinical samples. Lancet Infect Dis. 2020;20(4):411‐412. doi:10.1016/S1473-3099(20)30113-4. Available: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7128099/ (accessed 2020 May 25).

    [3] He X, Lau EHY, Wu P, et al. Temporal dynamics in viral shedding and transmissibility of COVID-19. Nat Med. 2020;26(5):672‐675. doi:10.1038/s41591-020-0869-5. Available: https://www.nature.com/articles/s41591-020-0869-5 (accessed 2020 May 25).

    Manohar R Furtado17 May 2020
    Limits of Detetction
    What was the Limit of Detection of the PCR tests used in these studies.While recommending "more sensitive tests" what should the Limits of Detection of a test be ?
    HJ Sievert13 May 2020
    Influence of viral concentration on false negatives
    To what extent would the viral concentration which, one would expect, increases from infection to the time when symptoms manifest themselves and then eventually decreases, result in false negatives given the typical sensitivity of the PCR test of around 100 viral copies/mL?