Rapid identification and isolation of infected individuals is crucial. Diagnosis is made using clinical, laboratory and radiological features. As symptoms and radiological findings of COVID-19 are non-specific, SARS-CoV-2 infection has to be confirmed by nucleic acid-based polymerase chain reaction (PCR), amplifying a specific genetic sequence in the virus. Within just a few days after the first cases were published, a validated diagnostic workflow for SARS-CoV-2 was presented (Corman 2020), demonstrating the enormous response capacity achieved through coordination of academic and public laboratories in national and European research networks.
There is an interim guidance for diagnostic testing for COVID-19 in suspected human cases, published by WHO in March and updated on September 11, 2020 (WHO 20200911). Several comprehensive up-to-date reviews of laboratory techniques in diagnosing SARS-CoV-2 have been published recently (Kilic 2020, Loeffelholz 2020).
Open the references of this chapter in a separate window.
According to WHO, the decision to test “should be based on both clinical and epidemiological factors”, in order to support clinical management of patients and infection control measures. In symptomatic patients, a PCR test should be immediately carried out, especially for medical professionals with symptoms. In particular, this applies to nursing homes and other long-term facilities where large outbreaks with high resident mortalty may occur. In these settings, every day counts: both residents and health-care workers should be tested immediately. In regression analyses among 88 nursing homes with a documented case before facility-wide testing occurred, each additional day between identification of the first case and completion of facility-wide testing was associated with identification of 1.3 additional cases (Hatfield 2020). However, the predictive value of the tests markedly varies with time from exposure and symptom onset. The false-negative rate is lowest 3 days after onset of symptoms, or approximately 8 days after exposure (see below).
In settings with limited resources, however, patients should only be tested if a positive test results in imperative action. It does not necessarily make sense to attempt to ascertain the prevalence of infection by PCR. For example, in a family which was put on quarantine after the infection was confirmed in one member, not all household contacts have to be tested, especially younger persons with only mild symptoms.
For many countries and regions, there are constantly updated recommendations by authorities and institutions about who should be tested by whom and when: these recommendations are constantly changing and have to be adapted to the local epidemiological situation. The lower the infection rates and the higher the testing capacities, the more patients will be able to be tested.
COVID Reference resources:
Home | Daily Top 10 | 7-Day Overview | Download | Epidemiology | Transmission | Prevention | Virology | Vaccines | Diagnosis | Clinical Manifestations | Treatment | Severe COVID-19 | Comorbidities / Special Populations | Pediatrics| Timeline | Preface
SARS-CoV-2 can be detected in a wide range of different tissues and body fluids. In a study on 1,070 specimens collected from 205 patients with COVID-19 (Wang X 2020), bronchoalveolar lavage fluid specimens showed the highest positive rates (14 of 15; 93%), followed by sputum (72 of 104; 72%), nasal swabs (5 of 8; 63%), fibrobronchoscopy brush biopsy (6 of 13; 46%), pharyngeal swabs (126 of 398; 32%), feces (44 of 153; 29%), and blood (3 of 307; 1%).
Though respiratory secretions may be quite variable in composition, respiratory samples remain the sample type of choice for diagnostics. Viral replication of SARS-CoV-2 is very high in upper respiratory tract tissues which is in contrast to SARS-CoV (Wolfel 2020). According to WHO, respiratory material for PCR should be collected from upper respiratory specimens (nasopharyngeal and oropharyngeal swab or wash) in ambulatory patients (WHO 2020). It is preferred to collect specimens from both nasopharyngeal and oropharyngeal swabs which can be combined in the same tube. Besides nasopharyngeal swabs, samples can be taken from sputum (if producible), endotracheal aspirate, or bronchoalveolar lavage. It is likely that lower respiratory samples are more sensitive than nasopharyngeal swabs. Especially in seriously ill patients, there is often more virus in the lower than in the upper respiratory tract (Huang 2020). However, there is always a high risk of “aerosolization” and thus the risk that staff members become infected.
A prospective study in two regional hospitals in Hong Kong examined 563 serial samples collected during the viral shedding period of 50 patients: 150 deep throat saliva (DTS), 309 pooled nasopharyngeal (NP) and throat swabs, and 104 sputum (instructions for deep throat saliva: first clear your throat by gargling with your own saliva, and then spit out the DTS into a sterile bottle). Deep throat saliva produced the lowest viral RNA concentration and a lower RT-PCR positive rate compared to conventional respiratory specimens. Buccal swabs do not work well either. In 11 children positive via nasopharyngeal swabs, 2 remained negative via buccal swabs. There was a general trend for buccal specimens to contain lower SARS-CoV-2 viral loads compared with nasopharyngeal specimens (Kam 2020).
Nasopharyngeal swabs – practical issues
It is important to carry out the swab process correctly. Both nasopharynx and oropharyngeal swabs have a number of error options that all can lead to false negative results. In addition, protective measures must be taken in order not to endanger the examiner. Every swab carries a high risk of infection! Respiratory protection, protective glasses, gowns and gloves are required. The correct putting on and taking off of protective clothing should be practiced! Many mistakes occur even just removing the protective mask. Gathering specimens from nasopharyngeal and throat swabs can cause discomfort for patients and put health-care workers at risk. If not performed properly or in patients with complex and delicate anatomy, there is a risk for adverse events such as cerebrospinal fluid leak (Sullivan 2020). There is a very useful video on protection, preparation, equipment, handling, removing personal protective equipment, etc (Marty 2020).
For the smear, the patient should sit on a chair and put his head slightly back. The examiner should stand at a slightly offset position in order to avoid any possible cough droplet. Tell the patient that it might be uncomfortable for a short time. Swabs should be used that are suitable for virus detection and have the most flexible plastic shaft possible. Wooden sticks can inactivate viruses and pose a high risk of injury. The swab should be held between thumb and forefinger, like a pencil, so the end should not touch anything. The posterior wall of the nasopharynx is often reached after 5-7 cm, indicated by a slight resistance. Mid-turbinate nasal swabs may be less sensitive (Pinninti 2020). Touching the teeth and tongue should be avoided when taking a throat swab; the swab should be removed from the back wall, directly next to the uvula. Caution with the gag reflex! There is a wealth of practical videos on the internet for the correct execution of the swab process.
In order to minimize the exposure risk to health care workers and depletion of personal protective equipment, we have established swab instructions for patients who are able to do this (ie, most of them!) at home. After appropriate instruction, they can perform the swabs themselves. A courier with the tubes is sent directly to the patient’s home, and the courier places the tubes at the door. Direct contact between patient and courier should be avoided. The swab tubes should not be touched by the courier (either put them directly in a bag or collect them with an inverted bag) and should be brought back directly (no mailing!). This requires prior, precise instruction, but is usually quite feasible. Unsupervised home swab collection was comparable to clinician-collected nasopharyngeal swab collection (McCulloch 2020). In one of the largest studies to date, a total of 530 patients with upper respiratory infection were provided with instructions and asked to collect tongue, nasal, and mid-turbinate samples (Tu 2020). A nasopharyngeal sample was then collected from the patient by a healthcare worker. When this NP sample was used as the comparator, the estimated sensitivities of the tongue, nasal, and mid-turbinate samples collected by the patients were 89.8%, 94.0% and 96.2%, respectively.
The swabs can be stored dry or in a small amount of NaCl solution; if necessary, this should be clarified with the laboratory beforehand. Quick PCR examination is important, preferably on the same day if possible. Heat and longer storage can lead to false negative results (Pan 2020).
Lower respiratory specimens may include sputum (if produced) and/or endotracheal aspirate or bronchoalveolar lavage in patients with more severe respiratory disease. However, a high risk of aerosolization should be considered (adhere strictly to infection prevention and control procedures). Additional clinical specimens may be collected as COVID-19 virus has been detected in blood and stools (see below).
In contrast to many respiratory viruses, SARS-CoV-2 is present in saliva and several studies have shown that posterior oropharyngeal (deep throat) saliva samples are feasible and more acceptable to patients and healthcare workers (To 2020, Yu 2020, Wyllie 2020, Yokota 2020). In a large study on “enhanced” saliva specimens (strong sniff, elicited cough, and collection of saliva/secretions) from 216 patients with symptoms deemed consistent with COVID-19, there was a 100% positive agreement (38/38 positive specimens) and 99.4% negative agreement (177/178 negative specimens).
Although no cases of transmission via fecal-oral route have yet been reported, there is also evidence that SARS-CoV-2 is actively replicating in the gastrointestinal tract. Several studies showed prolonged presence of SARS-CoV-2 viral RNA in fecal samples (Chen 2020, Wu 2020). Combining results of 26 studies, a rapid review revealed that 54% of those patients tested for fecal RNA were positive. Duration of fecal viral shedding ranged from 1 to 33 days after a negative nasopharyngeal swab (Gupta 2020). In another meta-analysis of 17 studies, the pooled detection rate of fecal SARS-CoV-2 RNA was 44% and 34% by patient and number of specimens, respectively. Patients who presented with gastrointestinal symptoms (77% vs. 58%) or with a more severe disease (68% vs. 35%) tended to have a higher detection rate.
These studies have raised concerns about whether patients with negative pharyngeal swabs are truly virus-free, or sampling of additional body sites is needed. However, the clinical relevance of these findings remains unclear and there is one study that did not detect infectious virus from stool samples, despite having high virus RNA concentrations (Wolfel 2020). Therefore, the presence of nucleic acid alone cannot be used to define viral shedding or infection potential (Atkinson 2020). For many viral diseases including SARS-CoV or MERS-CoV, it is well known that viral RNA can be detected long after the disappearance of infectious virus.
Specimens other than respiratory and gastrointestinal: blood, urine, breast milk
- Blood – in patients with mild or moderate disease, SARS-CoV-2 is relatively rarely detected in blood (Wang W 2020, Wolfel 2020). In a screening study of 7,425 blood donations in Wuhan, plasma samples were found positive for viral RNA from 2 asymptomatic donors (Chang 2020). Another study from Korea found seven asymptomatic blood donors who were later identified as COVID-19 confirmed cases. None of 9 recipients of platelets or red blood cell transfusions tested positive for SARS-CoV-2 RNA. Transfusion transmission of SARS-CoV-2 was considered to be unlikely (Kwon 2020). As with feces, it remains unclear whether detectable RNA in the blood signifies infectivity. In a study of 167 hospitalized patients, SARS-CoV-2 was found in 64 patients at hospital admission, 3 of 106 serum PCR negative patients and 15 of 61 positive patients died (Hagman 2020). However, the clinical significance of SARS-CoV-2 “RNAemia” needs to be defined.
- Urine – None of 72 urine specimens tested positive (Wang X 2020).
- Breast milk – in a case report, SARSCoV2 RNA was detected in breast milk samples from an infected mother on 4 consecutive days. Detection of viral RNA in milk coincided with mild COVID19 symptoms and a SARSCoV2 positive diagnostic test of the newborn (Groß 2020). However, this seems to be rare. Among 64 breast milk samples from 18 infected women, SARS-CoV-2 RNA was detected in only one milk sample; the viral culture for that sample was negative. These data suggest that SARS-CoV-2 RNA does not represent replication-competent virus and that breast milk may not be a source of infection for the infant (Chambers 2020. Case reports of transmitted antibodies in breast milk have also been reported (Dong 2020).
- Vaginal fluid – all samples of 10 women with COVID-19 were negative (Saito 2020).
- Semen – Absence of virus in samples collected from 12 patients in their recovery phase (Song 2020).
- Tears and conjunctival secretions – among 40 patients (10 with conjunctivitis) who tested positive by RT-PCR of nasopharyngeal and oropharyngeal swabs, conjunctival swab PCR was positive for 3 patients, among them one with conjunctivitis (Atum 2020).
Dozens of in-house and commercial rRT-PCR assays are available as labs worldwide have customized their PCR tests for SARS-CoV-2, using different primers targeting different sections of the virus’s genetic sequence. A review of different assays and diagnostic devices was recently published (Loeffelholz 2020). A protocol for real-time (RT)-PCR assays for the detection of SARS-CoV-2 for two RdRp targets (IP2 and IP4) is described at https://www.who.int/docs/default-source/coronaviruse/real-time-rt-pcr-assays-for-the-detection-of-sars-cov-2-institut-pasteur-paris.pdf?sfvrsn=3662fcb6_2
Novel real-time RT-PCR assays targeting the RNA-dependent RNA polymerase (RdRp)/helicase, spike and nucleocapsid genes of SARS-CoV-2 may help to improve the laboratory diagnosis of COVID-19. Compared to the reported RdRp-P2 assay which is used in most European laboratories, these assays do not cross-react with SARS-CoV in cell culture and may be more sensitive and specific (Chan JF 2020).
The limits of detection of commercial kits may differ substantially. However, most comparative studies have shown a high sensitivity and their suitability for screening purposes worldwide:
- In a comparison of 11 different RT-PCR test systems used in seven labs in Germany in March 2020, the majority of RT-PCR assays detected ca 5 RNA copies per reaction (Münchhoff 2020). A reduced sensitivity was noted for the original Charité RdRp gene confirmatory protocol, which may have impacted the confirmation of some cases in the early weeks of the pandemic. The CDC N1 primer/probe set was sensitive and robust for detection of SARS-CoV-2 in nucleic acid extracts from respiratory material, stool and serum from COVID-19 patients.
- Analytical limits of detection for seven SARS-CoV-2 assays using serial dilutions of pooled patient material quantified with droplet digital PCR. Limits of detection ranged from ≤ 10 to 74 copies/ml for commercial high-throughput laboratory analyzers (Roche cobas, Abbott m2000, and Hologic Panther Fusion) and 167 to 511 copies/ml for sample-to-answer (DiaSorin Simplexa, GenMark ePlex) and point-of-care instruments (Abbott ID NOW) (Fung 2020).
- A total of 239 specimens (168 contained SARS-CoV-2) were tested by five test methods (Procop 2020). The assays that lacked a nucleic acid extraction step produced more false-negative reactions than assays that included this step. The false-negative rates were 0% for the CDC 2019 nCoV Real-Time RT-PCR Diagnostic Panel, 3.5% for TIB MOLBIOL Assay (Roche), 2.4% for Xpert Xpress SARS-CoV-2 (Cepheid), 11.9% for Simplexa COVID-19 Direct Kit (DiaSorin), and 16.7% for the ID NOW COVID-19 (Abbott). Most false negatives were seen in patients with low viral loads.
A qualitative PCR (“positive or negative”) is usually sufficient in routine diagnostics. Quantification of viral RNA is currently (still) only of academic interest.
False positive results are very rare. However, they do occur. Though the analytical specificity of these tests is usually 100%, the clinical specificity is less, due to contamination (a significant problem for NAT procedures) and/or human error in the handling of samples or data (very hard to eliminate entirely). As seen with serology (see below), these false positive results can have substantial effects when prevalence is low (Andrew Cohen, personal communication).
Another problem of any qualitative PCR is false negative results which can have many causes (review: Woloshin 2020). Incorrect smears are particularly common, but laboratory errors also occur. In a review of 7 studies with a total of 1,330 respiratory samples, the authors estimated the false-negative rate of RT-PCR by day since infection. Over the 4 days before symptom onset, the rate decreased from 100% to 67%. On the day of symptom onset (day 5), the rate was 38%, decreasing to 20% (day 8) and then beginning to increase again from 21% (day 9) to 66% (day 21). If clinical suspicion is high, infection should not be ruled out on the basis of RT-PCR alone. The false-negative rate is lowest 3 days after onset of symptoms, or approximately 8 days after exposure (Kucirka 2020). Figure 1 illustrates PCR and antibody detection during SARS-CoV-2 infection.
Figure 1. Timeline of diagnostic markers for detection of SARS-CoV-2. AB = Antibody.
Do we need to re-test in the case of a negative PCR? Several studies argue against this strategy, finding very low rates of negative-to-positive conversion with repeated testing (Lepak 2020). Among 20,912 patients, one study analyzed the frequency of SARS-CoV-2 RT-PCR test discordance among individuals initially testing negative by nasopharyngeal swab who were retested on clinical grounds within 7 days. The frequency of subsequent positivity within this window was only 3.5% and similar across institutions (Long 2020). It appears that if the first PCR is negative, a second PCR only yields a small number of positive results.
Several studies have shown that asymptomatic patients also have positive PCR results and can transmit the virus (Bai 2020, Cereda 2020, Rothe 2020). The cycle threshold values of RT-PCR for SARS-CoV-2 (“viral load”) in asymptomatic patients are similar to those in symptomatic patients (Lee S 2020, Lavezzo 2020).
In symptomatic patients, viral shedding may begin 2 to 3 days before the appearance of the first symptoms. Analyzing a total of 414 throat swabs in 94 patients, the highest viral load in throat swabs was found at the time of symptom onset. Infectiousness started from 2.3 days (95% CI, 0.8–3.0 days) before symptom onset and peaked at 0.7 days before symptom onset (He 2020). Infectiousness was estimated to decline quickly within 7 days.
In a cohort of 113 symptomatic patients, the median duration of detection of SARS-CoV-2 RNA was 17 days (interquartiles 13-22 days), measured from the onset of the disease. In some patients, PCR was positive even longer: male gender and a severe course (invasive mechanical ventilation) were independent risk factors for prolonged shedding (Xu K 2020).
Several reports from patients have repeatedly gained much media attraction, showing positive results after repeated negative PCR and clinical recovery (Lan 2020, Xiao AT 2020, Yuan 2020). These studies have raised the question of re-activation or re-infection of COVID-19 (see below, chapter Clinical Presentation, page 279). However, it seems probable that the results are much more likely due to methodological problems (Li 2020). At low virus levels, especially during the final days of infection, the viral load can fluctuate and sometimes be detectable, sometimes not (Wolfel 2020). Reactivation, and also a rapid reinfection would be very unusual for coronaviruses.
Quantification of viral load
Several studies have evaluated the SARS-CoV-2 viral load in different specimens. In a small prospective study, the viral load in nasal and throat swabs obtained from 17 symptomatic patients was analyzed in relation to day-of-onset of any symptoms (Zou 2020). Of note, the viral load detected in asymptomatic patients was similar to that in symptomatic patients, which suggests the transmission potential of asymptomatic or minimally symptomatic patients.
In another study on 82 infected individuals, the viral loads in throat swab and sputum samples peaked at around 5–6 days after symptom onset, ranging from around 79,900 copies/ml in the throat to 752,000 copies per mL in sputum (Pan 2020). In a study on oropharyngeal saliva samples, unlike SARS, patients with COVID-19 had the highest viral load near presentation, which could account for the fast-spreading nature of this epidemic (To 2020). The median viral load in posterior oropharyngeal saliva or other respiratory specimens at presentation was 5.2 log10 copies per mL (IQR 4.1-7.0) in this study. In a total of 323 samples from 76 patients, the average viral load in sputum (17,429 copies/test) was significantly higher than in throat swabs (2,552 copies) and nasal swabs (651 copies). Viral load was higher in the early and progressive stages than in the recovery stage (Yu 2020). According to a recently published study, viral shedding may already begin 2-3 days before the appearance of the first symptoms and the infectiousness profile may more closely resemble that of influenza than of SARS (He 2020).
Higher viral loads might be associated with severe clinical outcomes. In a large cohort (n = 1145) of hospitalized, symptomatic patients from New York, viral loads were measured. In a Cox proportional hazards model adjusting for several confounders, there was a significant independent association between viral load and mortality (hazard ratio 1.07, 95% CI 1.03–1.11, p = 0.0014), with a 7% increase in hazard for each log transformed copy/mL (Pujadas 2020). However, prospective trials are needed to evaluate the role of SARS-CoV-2 viral load as a marker for assessing disease severity and prognosis.
Should we measure viral load? Probably yes. It may be helpful in clinical practice. A positive RT-qPCR result may not necessarily mean the person is still infectious or that they still have any meaningful disease. The RNA could be from non-viable virus and/or the amount of live virus may be too low for transmission.
Cycle threshold (Ct) values
RT-qPCR provides quantification by first reverse transcribing RNA into DNA, and then performing qPCR where a fluorescence signal increases proportionally to the amount of amplified nucleic acid. The test is positive if the fluorescence reaches a specified threshold within a certain number of PCR cycles (Ct value, inversely related to the viral load). Many qPCR assays use a Ct cut-off of 40, allowing detection of very few starting RNA molecules. Some experts (Tom 2020) suggest using this Ct value or to calculate viral load which can help refine decision-making (shorter isolation, etc). Unfortunately, there is still a wide heterogeneity and inconsistency of the standard curves calculated from studies that provide Ct values from serial dilution samples and the estimated viral loads. According to other experts, precautions are needed when interpreting the Ct values of SARS-CoV-2 RT-PCR results shown in COVID-19 publications to avoid misunderstanding of viral load kinetics for comparison across different studies (Han 2020). Caution is needed when regarding Ct values as a surrogate indicator of ‘quantity’ in a qualitative PCR assay (“viral load”). Results are not transferable across different assays, different gene targets and different specimen types (Poon 2020).
However, some clinical key studies are listed here:
- In 678 patients with COVID-19, in-hospital mortality was 35.0% with a “high viral load” (Ct < 25; n = 220), 17.6% with a “medium viral load” (Ct 25-30; n = 216), and 6.2% with a “low viral load” (Ct > 30; n = 242). High viral load was independently associated with mortality (adjusted odds ratio 6.05; 95% CI: 2.92-12.52) and intubation (aOR 2.73; 95% CI: 1.68-4.44) in multivariate models (Magleby 2020).
- A prospective serial sampling of 70 patients revealed clinically relevant Ct values, namely a Ct of 24 (“high viral load”), and > 40 (“negative”), occurred 9 and 36 days after symptom onset (Lesho 2020).
- Among 93 household members (including index cases) who tested positive for SARS-CoV-2 by NP swab, Ct values were lowest soon after symptom onset and were significantly correlated with time elapsed since onset; within 7 days after symptom onset, the median Ct value was 26.5, compared with a median Ct value of 35.0 at 21 days after onset (Salvatore 2020).
- Virus culture was attempted from 324 samples (from 253 cases) that tested positive for SARS-CoV-2 by RT-PCR. Ct values correlated strongly with cultivable virus. Probability of culturing virus declined to 8% in samples with Ct > 35 and to 6% (95% CI: 0.9–31.2%) 10 days after onset (Singanayagam 2020).
- A cross-sectional study determined PCR positive samples for their ability to infect cell lines. Of 90 samples, only 29% demonstrated viral growth. There was no growth in samples with a Ct > 24 or duration of symptoms > 8 days (Bullard 2020).
Test systems other than conventional RT-PCR
Access to rapid diagnosis is key to the control of the SARS-CoV-2 pandemic. In the future, point-of-care testing could relieve pressure on centralized laboratories and increase overall testing capacity. Besides PCR, additional potentially valuable amplification/detection methods, such as CRISPR (targeting clustered regularly interspaced short palindromic repeats), isothermal nucleic acid amplification technologies (e.g. reverse transcription loop-mediated isothermal amplification (RT-LAMP), and molecular microarray assays are under development or are in the process of being commercialized. According to WHO on September 11, validation of the analytic and clinical performance of these assays, demonstration of their potential operational utility, rapid sharing of data, as well as emergency regulatory review of manufacturable, well-performing tests “are encouraged to increase access to SARS-CoV-2 testing” (WHO 20200911).
Point-of-care tests are easy-to-use devices to facilitate testing outside of laboratory settings (Guglielmi 2020, Joung 2020). They are eagerly awaited. But will they be game-changers? On May 6, the FDA granted an emergency use authorization for a CRISPR-based SARS-CoV-2 fluorescent assay marketed by Sherlock Biosciences. This straightforward SARS-CoV-2 test combines simplified extraction of viral RNA with isothermal amplification and CRISPR-mediated detection. The results are available within an hour with minimal equipment. First results (n = 202 positive/200 negative samples): sensitivity 93.1%, specificity 98.5% (Joung 2020). However, its use still remains limited to laboratories certified to perform high-complexity tests. There are other reports of an all-in-one dual CRISPR-Cas12a assay (Ding 2020) which allows all components to be incubated in one pot for CRISPR-based nucleic acid detection, enabling simple, all-in-one molecular diagnostics without the need for separate and complex manual operations.
On May 6, FDA also authorized (EUA) Quidel’s Sofia 2 SARS Antigen Fluorescent Immunoassay. This test must be read on a dedicated analyzer and detects SARS-CoV-2 nucleocapsid protein from nasopharyngeal swabs in 15 min. According to the manufacturer, the assay demonstrated acceptable clinical sensitivity and detected 47/59 infections (80%). In another study, the so called CovidNudge test had 94% sensitivity and 100% specificity when compared with standard laboratory-based RT-PCR (Gibani 2020). In other studies, sensitivity was much lower. The BIOCREDIT COVID-19 antigen test was 10,000 fold less sensitive than RT-PCR and detected between 11.1 % and 45.7% of RT-PCR-positive samples from COVID-19 patients (Mak 2020).
Besides antigen tests, several rapid nucleic acid amplification tests have been recently released (Collier 2020). The Abbott ID NOW COVID-19 assay (using isothermal nucleic acid amplification of the RdRp viral target) is capable of producing positive results in as little as 5 minutes. In one stuy, results were compared with RT-PCR Cepheid Xpert Xpress SARS-CoV-2 using nasopharyngeal swabs (Basu 2020). Regardless of method of collection and sample type, the rapid test had negative results in a third of the samples that tested positive by PCR when using nasopharyngeal swabs in viral transport media and 45% when using dry nasal swabs. Such “Reverse Transcription Loop-Mediated Isothermal Amplification” tests (RT-LAMP) could be used outside of a central laboratory on various types of biological samples. They can be completed by individuals without specialty training or equipment and may provide a new diagnostic strategy for combating the spread of SARS-CoV-2 at the point-of-risk (Lamb 2020).
Given the low (or still unproven) sensitivity, these tests may mainly serve as an early adjunctive tool to identify infectious individuals very rapidly, i.e. in the emergency unit. These tests help to avoid bed closure, allow discharge to care homes and expedite access to hospital procedures. Some experts are even more optimistic: the frequent use of cheap, simple, rapid tests is essential, even if their analytic sensitivities are vastly inferior to those of benchmark tests. The key question is not how well molecules can be detected in a single sample – but how effectively infections can be detected in a population by the repeated use of a given test as part of an overall testing strategy – the sensitivity of the testing regimen (Mina 2020).
Diagnosis in the setting of a shortage of PCR test kits
There is no doubt that the overall goal must be to detect as many infections as possible. However, in many countries, a shortage of supply test kits does not meet the needs of a growing infected population. Especially in low-prevalence settings, sample pooling is an option to reduce costs and speed results. In this approach, small volumes of samples from multiple patients are combined into a single test, resulting in substantial reagent savings. Several studies have shown that 5-10 samples can be pooled, without compromising the results (Ben-Ami 2020, Schmidt 2020). However, pooling is not that trivial (Mallapaty 2020). There are several caveats and careful and rigorous investigation is necessary to assure that the pooling of specimens does not impact the analytical sensitivity of the assay (review: Clark 2020).
Some studies have investigated whether the diagnosis in high prevalence periods and countries can be made without PCR detection if necessary. A large retrospective case-control study from Singapore has evaluated predictors for SARS-CoV-2 infection, using exposure risk factors, demographic variables, clinical findings and clinical test results (Sun 2020). Even in the absence of exposure risk factors and/or radiologic evidence of pneumonia, clinical findings and tests can identify subjects at high risk of COVID-19. Low leukocytes, low lymphocytes, higher body temperature, higher respiratory rate, gastrointestinal symptoms and decreased sputum production were strongly associated with a positive SARS-CoV-2 test. However, those preliminary prediction models are sensitive to the local epidemiological context and phase of the global outbreak. They only make sense during times of high incidence. In other words: if I see a patient during the peak of an epidemic presenting with fever, cough, shortness of breath and lymphopenia, I can be almost sure that this patient suffers from COVID-19. During phases when the incidence is lower, these models do not make sense. There is no doubt that the nucleic acid test serves as the gold standard method for confirmation of infection. Whenever PCR is available, PCR should be performed.
Serology (antibody testing)
Detection of past viral infections by looking for antibodies an infected person has produced will be among the most important goals in the fight against the COVID-19 pandemic. Antibody testing is multipurpose: these serological assays are of critical importance to determine seroprevalence, previous exposure and identify highly reactive human donors for the generation of convalescent serum as therapeutic. They will support contact tracing and screening of health care workers to identify those who are already immune. How many people really got infected, in how many did the virus escape the PCR diagnosis, and for what reasons, how many patients are asymptomatic, and what is the real mortality rate in a defined population? Only with comprehensive serology testing (and well-planned epidemiological studies) will we be able to answer these questions and reduce the ubiquitous undisclosed number in the current calculations. Several investigations are already underway in a wide variety of locations worldwide.
In recent weeks it has become clear that serology testing may also aid as a complementary diagnostic tool for COVID-19. The seroconversion of specific IgM and IgG antibodies were observed as early as the 4th day after symptom onset. Antibodies can be detected in the middle and later stages of the illness (Guo L 2020, Xiao DAT 2020). If a person with a highly suspicious COVID-19 remains negative by PCR testing and if symptoms are ongoing for at least several days, antibodies may be helpful and enhance diagnostic sensitivity.
However, antibody testing is not trivial. The molecular heterogeneity of SARS-CoV-2 subtypes, imperfect performance of available tests and cross-reactivity with seasonal CoVs have to be considered (reviews: Cheng 2020, Krammer 2020). According to a Cochrane analysis on 57 publications with 15,976 samples, the sensitivity of antibody tests is too low in the first week from symptom onset to have a primary role in the diagnosis of COVID-19. However, these tests may still have a role in complementing other testing in individuals presenting later, when RT-PCR tests are negative or are not done (Deeks 2020). Antibody tests are likely to have a useful role in detecting previous SARS-CoV-2 infection if used 15 or more days after the onset of symptoms. Data beyond 35 days post-symptom onset is scarce. According to the authors, studies of the accuracy of COVID-19 tests require considerable improvement. Studies must report data on sensitivity disaggregated by time from onset of symptoms. A practical overview of the pitfalls of antibody testing and how to communicate risk and uncertainty is given by Watson 2020.
Average sensitivity and specificity of FDA-approved antibody tests is 84.9% and 98.6%, respectively. Given variable prevalence of COVID-19 (1%-15%) in different parts, statistically the positive predictive value will be as low as 30% to 50% in areas with low prevalence (Mathur 2020). A systematic review of 40 studies on sensitivity and specificity was recently published (Bastos 2020), stratified by method of serological testing (enzyme linked immunosorbent assays – ELISAs), lateral flow immunoassays (LFIAs), or chemiluminescent immunoassays – CLIAs). The pooled sensitivity of ELISAs measuring IgG or IgM was 84.3% (95% confidence interval 75.6% to 90.9%), of LFIAs was 66.0% (49.3% to 79.3%), and of CLIAs was 97.8% (46.2% to 100%). According to the authors, higher quality clinical studies assessing the diagnostic accuracy of serological tests for COVID-19 are urgently needed.
A nice overview of the different platforms, including binding assays such as enzyme-linked immunosorbent assays (ELISAs), lateral flow assays, or Western blot–based assays is given by Krammer 2020. In addition, functional assays that test for virus neutralization, enzyme inhibition, or bactericidal assays can also inform on antibody-mediated immune responses. Many caveats and open questions with regard to antibody testing are also discussed.
Antibody testing usually focuses on antigens (proteins). In the case of SARS-CoV-2, different ELISA kits based on recombinant nucleocapsid protein and spike protein are used (Loeffelholz 2020). The SARS-CoV-2 spike protein seems to be the best target. However, which part of the spike protein to use is less obvious and there is a lot hanging on the uniqueness of the spike protein. The more unique it is, the lower the odds of cross-reactivity with other coronaviruses—false positives resulting from immunity to other coronaviruses. Cross reactivity to other coronaviruses can be challenging. So called confirmation tests (usually neutralization tests) can be used to reduce false positive testings. However, detection and quantification of neutralizing antibodies are relatively low-throughput and limited to Biosafety Level 3-equipped research laboratories. To avoid neutralization tests that require live pathogen and a biosafety level 3 laboratory, several studies have proposed tests based on antibody-mediated blockage of the interaction between the ACE2 receptor protein and the receptor-binding domain. The tests achieved 99.93% specificity and 95–100% sensitivity (Tan 2020).
Even with a very high specificity of 99% and above, however, especially in low-prevalence areas, the informative value of antibody testing is limited and a high rate of false positive tests can be assumed. An example: With a specificity of 99%, it is expected that one test out of 100 is positive. In a high prevalence setting, this is less relevant. However, if a person is tested in a low prevalence setting, the likelihood that a positive test is really positive (the positive predictive value, i.e. the number of really positive tests divided by the number of all positive tests) is low. In a population with a given prevalence of 1%, the predictive value would only be 50%! Current estimates from Iceland, a well-defined but unselected population, still have shown a relatively constant rate of around 0.8% in March 2020 (Gudbjartsson 2020). Even in apparently more severely affected countries, the infection rates are only slightly higher. General antibody screening in these populations will therefore produce a fairly high rate of false positive tests. When assessing anti-SARS-CoV-2 immune status in individuals with low pre-test probability, it may be better to confirm positive results from single measurements by alternative serology tests or functional assays (Behrens 2020).
Some key studies with head-to-head-assessments of different immunoassays
- Abbott, EUROIMMUN and the Elecsys (Roche): The Abbott assay demonstrated the fewest false negative results > 14d post-symptom onset and the fewest false positive results. While the Roche assay detected more positive results earlier after onset of symptoms, none of the assays demonstrated high enough clinical sensitivity before day 14 from symptom onset to diagnose acute infection (Tang 2020).
- Abbott, LIAISON (DiaSorin), Elecsys (Roche), Siemens, plus a novel in-house 384-well (Oxford) ELISA in 976 (!) pre-pandemic blood samples and 536 (!) blood samples with confirmed SARS-CoV-2 infection. All assays had a high sensitivity (92.7-99.1%) and specificity (98.7-99.9%). The most sensitive test assessed was the in-house ELISA. The Siemens assay and Oxford immunoassay achieved 98% sensitivity/specificity without further optimization. However, a limitation was the small number of pauci-symptomatic and asymptomatic cases analyzed (NAEG 2020).
- Abbott, Epitope Diagnostics, EUROIMMUN, and Ortho Clinical Diagnostics: all four immunoassays performed similarly with respect to sensitivity in COVID-19 hospitalized patients, and except for the Epitope assay, also in individuals with milder forms of the infection (Theel 2020). The Abbott and Ortho Clinical immunoassays provided the highest overall specificity, of over 99%.
Indication in clinical practice
But outside clinical studies, who should be tested now? Testing actually makes no sense for patients with a previous, proven COVID-19 disease. However, it can still be done if, for example, you want to validate a test. In addition to those involved in health care or working in other professions with a high risk of transmission, such testing can also be useful in order to identify possible contact persons retrospectively. However, we only measure antibodies when the testing result might have consequences. Patients should be informed about the low positive predictive value, especially in those without any evidence of prior disease or exposition to COVID-19. In these patients, antibody testing is not recommended. Outside epidemiological hot spots, in low prevalence countries like Germany, virtually everybody is still seronegative. If positive, the predictive value is too low.
The kinetics of antibodies
Serologic responses to coronaviruses are only transient. A brilliant systematic review of antibody-mediated immunity to coronaviruses (kinetics, correlates of protection, and association with severity) was recently published (Huang AT 2020).
Antibodies to other human, seasonal coronaviruses may disappear even after a few months. Preliminary data suggest that the profile of antibodies to SARS-CoV-2 is similar to SARS-CoV (Xiao DAT 2020). For SARS-CoV, antibodies were not detected within the first 7 days of illness, but IgG titre increased dramatically on day 15, reaching a peak on day 60, and remained high until day 180 from when it declined gradually until day 720. IgM was detected on day 15 and rapidly reached a peak, then declined gradually until it was undetectable on day 180 (Mo 2006). As with other viruses, IgM antibodies occur somewhat earlier than IgG antibodies which are more specific. IgA antibodies are relatively sensitive but less specific (Okba 2020).
The first larger study on the host humoral response against SARS-CoV-2 has shown that these tests can aid the diagnosis of COVID-19, including subclinical cases (Guo 2020). In this study, IgA, IgM and IgG response using an ELISA-based assay on the recombinant viral nucleocapsid protein was analyzed in 208 plasma samples from 82 confirmed and 58 probable cases (Guo 2020). The median duration of IgM and IgA antibody detection were 5 days (IQR 3-6), while IgG was detected on day 14 (IQR 10-18) after symptom onset, with a positive rate of 85%, 93% and 78% respectively. The detection efficiency by IgM ELISA was higher than that of PCR after 5.5 days of onset of symptoms. In another study of 173 patients, the seroconversion rates (median time) for IgM and IgG were 83% (12 days) and 65% (14 days), respectively. A higher titer of antibodies was independently associated with severe disease (Zhao 2020). In other studies, however, antibody level did not correlate clearly with clinical outcomes (Ren 2020).
In some patients, IgG occurs even faster than IgM. In a study on seroconversion patterns of IgM and IgG antibodies, the seroconversion time of IgG antibody was earlier than IgM. IgG antibody reached the highest concentration on day 30, while IgM antibody peaked on day 18, but then began to decline (Qu J 2020). The largest study to date reported on acute antibody responses in 285 patients (mostly non-severe COVID-19). Within 19 days after symptom onset, 100% of patients tested positive for antiviral IgG. Seroconversion for IgG and IgM occurred simultaneously or sequentially. Both IgG and IgM titers plateaued within 6 days after seroconversion. The median day of seroconversion for both IgG and IgM was 13 days post-symptom onset. No association between plateau IgG levels and clinical characteristics was found (Long 2020).
However, there is some evidence that asymptomatic individuals develop less strong antibody responses. Moreover, antibodies disappear from the blood. Your COVID pass expires within a few weeks. Compared to symptomatic patients, 37 asymptomatic patients had lower virus-specific IgG levels in the acute phase (Long Q 2020). IgG levels and neutralizing antibodies started to decrease within 2–3 months after infection. Of note, 40% became seronegative (13% of the symptomatic group) for IgG in the early convalescent phase. Among 19 health care workers who had anti–SARS-CoV-2 antibodies detected at baseline, only 8 (42%) had antibodies that persisted above the seropositivity threshold at 60 days, whereas 11 (58%) became seronegative (Patel 2020). A decrease in anti-RBD antibody level was also seen in 15 donors of convalescent plasma (Perreault 2020).
Taken together, antibody testing is not only an epidemiological tool. It may also help in diagnosis. It will be seen in the coming months how the human antibody response to SARS-CoV-2 evolves over time and how this response and titres correlate with immunity. It is also conceivable that in some patients (e.g. those with immunodeficiency) the antibody response remains reduced.
Chest computed tomography
Computed tomography (CT) can play a role in both diagnosis and assessment of disease extent and follow-up. Chest CT has a relatively high sensitivity for diagnosis of COVID-19 (Ai 2020, Fang 2020). However, around half of patients may have a normal CT during the first 1-2 days after symptom onset (Bernheim 2020). On the other hand, it became clear very early in the current pandemic that a considerable proportion of subclinical patients (scans done before symptom onset) may already have pathological CT findings (Chan 2020, Shi 2020). In some of these patients showing pathological CT findings evident for pneumonia, PCR in nasopharyngeal swabs was still negative (Xu 2020). On the other hand, half of the patients who later develop CT morphologically visible pneumonia can still have a normal CT in the first 1-2 days after the symptoms appear (Bernheim 2020).
However, one should not overestimate the value of chest CT. The recommendation by some Chinese researchers to include CT as an integral part in the diagnosis of COVID-19 has led to harsh criticism, especially from experts in Western countries. The Chinese studies have shown significant errors and shortcomings. In view of the high effort and also due to the risk of infection for the staff, many experts strictly reject the general CT screening in SARS-CoV-2 infected patients or in those suspected cases (Hope 2020, Raptis 2020). According to the recommendation of the British Radiology Society, which made attempts to incorporate CT into diagnostic algorithms for COVID-19 diagnostics, the value of CT remains unclear – even if a PCR is negative or not available (Nair 2020, Rodrigues 2020). A chest CT should only be performed if complications or differential diagnoses are considered (Raptis 2020).
In blinded studies, radiologists from China and the United States have attempted to differentiate COVID-19 pneumonia from other viral pneumonia. The specificity was quite high, the sensitivity nuch lower (Bai 2020). A recent metaanalysis found a high sensitivity but low specificity (Kim 2020). The sensitivity of CT was affected by the distribution of disease severity, the proportion of patients with comorbidities, and the proportion of asymptomatic patients. In areas with low prevalence, chest CT had a low positive predictive value (1.5-30.7%).
If pathological, images usually show bilateral involvement, with multiple patchy or ground-glass opacities (GGO) with subpleural distribution in multiple bilateral lobes. Lesions may display significant overlap with those of SARS and MERS (Hosseiny 2020). According to a review of 45 studies comprising 4410 (!) patients, ground glass opacities (GGOs), whether isolated (50%) or co-existing with consolidations (44%) in bilateral and subpleural distribution, were the most prevalent chest CT findings (Ojha 4410). Another systematic review of imaging findings in 919 patients found bilateral multilobar GGO with a peripheral or posterior distribution, mainly in the lower lobes and less frequently within the right middle lobe as the most common feature (Salehi 2020). In this review, atypical initial imaging presentation of consolidative opacities superimposed on GGO were found in a smaller number of cases, mainly in the elderly population. Septal thickening, bronchiectasis, pleural thickening, and subpleural involvement were less common, mainly in the later stages of the disease. Pleural effusion, pericardial effusion, lymphadenopathy, cavitation, CT halo sign, and pneumothorax were uncommon (Salehi 2020).
The evolution of the disease on CT is not well understood. However, with a longer time after the onset of symptoms, CT findings are more frequent, including consolidation, bilateral and peripheral disease, greater total lung involvement, linear opacities, “crazy-paving” pattern and the “reverse halo” sign (Bernheim 2020). Some experts have proposed that imaging can be sorted into four different phases (Li M 2020). In the early phase, multiple small patchy shadows and interstitial changes emerge. In the progressive phase, the lesions increase and enlarge, developing into multiple GGOs as well as infiltrating consolidation in both lungs. In the severe phase, massive pulmonary consolidations and “white lungs” are seen, but pleural effusion is rare. In the dissipative phase, the GGOs and pulmonary consolidations were completely absorbed, and the lesions began to change into fibrosis.
In a longitudinal study analyzing 366 serial CT scans in 90 patients with COVID-19 pneumonia, the extent of lung abnormalities progressed rapidly and peaked during illness days 6-11 (Wang Y 2020). The predominant pattern of abnormalities after symptom onset in this study was ground-glass opacity (45-62%). As pneumonia progresses, areas of lesions enlarge and developed into diffuse consolidations in both lungs within a few days (Guan 2020).
Most patients discharged had residual disease on final CT scans (Wang Y 2020). Studies with longer follow-up are needed to evaluate long-term or permanent lung damage including fibrosis, as is seen with SARS and MERS infections. Pulmonary fibrosis is expected to be the main factor leading to pulmonary dysfunction and decline of quality of life in COVID-19 survivors after recovery. More research is needed into the correlation of CT findings with clinical severity and progression, the predictive value of baseline CT or temporal changes for disease outcome, and the sequelae of acute lung injury induced by COVID-19 (Lee 2020).
Of note, chest CT is not recommended in all COVID-19 patients, especially in those who are well enough to be sent home or those with only short symptomatic times (< 2 days). In the case of COVID-19, a large number of patients with infection or suspected infection swarm into the hospital. Consequently, the examination workload of the radiology department increases sharply. Because the transmission route of SARS-CoV-2 is through respiratory droplets and close contact transmission, unnecessary CT scan should be avoided. An overview of the prevention and control of the COVID-19 epidemic in the radiology department is given by An et al.
Ultrasound, PET and other techniques
Some experts have postulated that lung ultrasound (LUS) may be helpful, since it can allow the concomitant execution of clinical examination and lung imaging at the bedside by the same doctor (Buonsenso 2020, Soldati 2020). Potential advantages of LUS include portability, bedside evaluation, safety and possibility of repeating the examination during follow-up. Experience especially from Italy with lung ultrasound as a bedside tool has improved evaluation of lung involvement, and may also reduce the use of chest x-rays and CT. A point scoring system is employed by region and ultrasound pattern (Vetrugno 2020). However, the diagnostic and prognostic role of LUS in COVID-19 is uncertain.
In patients with neurological symptoms, brain MRI is often performed. In 27 patients, the most common imaging finding was cortical signal abnormalities on FLAIR images (37%), accompanied by cortical diffusion restriction or leptomeningeal enhancement (Kandemirli 2020). However, the complex clinical course including comorbidities, long ICU stay with multidrug regimens, respiratory distress with hypoxia episodes can all act as confounding factors and a clear cause-effect relationship between COVID-19 infection and MRI findings will be hard to establish.