Healthcare workers are at increased risk for severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) infection resulting in severe coronavirus disease 2019 (COVID-19) [1
]. Properly used protection equipment can reduce transmission risk but direct patient contact, endotracheal intubation and contact with contagious body fluids are associated with an increased infection risk [4
]. In turn, infected healthcare workers pose a significant threat to patients they care for [5
]. People with a compromised immune system or on treatment with immunosuppressive drugs, such as patients with chronic kidney disease (CKD) including those on dialysis treatment or with a kidney transplant, are among the most vulnerable with respect to life-threatening infectious diseases [6
]. Regardless of the “stay at home-stay safe” practice during the COVID-19 pandemic, they are in need of nondeferrable admission to kidney centers.
Reports on SARS-CoV‑2 infections among patients with CKD showed a mortality of up to 28% in kidney transplant recipients or solid organ transplants [9
]. Early studies from China revealed a surprisingly low mortality in dialysis patients, which contrasts with reports from the Austrian Dialysis and Transplant registry [12
]. On 8 May 2020 the COVID-19-specific mortality was 27% (12/44), which is comparable to the reported rate of 31% (18/59) in a recent report from the Columbia University Irving Medical Center, New York [14
]. It is well established that containment strategies in Austria were successful in preventing a collapse of the acute care facilities. More importantly, COVID-19-specific mortality was low as compared to other European countries [16
While reverse transcriptase polymerase chain reaction (RT-PCR) amplification of SARS-CoV‑2 RNA has its utility to identify acutely infected patients, serology testing is important to identify patients that have been infected in the past. Thus, detection of SARS-CoV‑2 specific antibodies is a prevalence marker in a population and can be used to measure herd immunity [18
]. In patients suffering from COVID-19 with RT-PCR proven SARS-CoV‑2 infection up to 100% tested positive for antiviral immunoglobulin G (IgG) within 3 weeks after symptom onset. Seroconversion for IgG and immunoglobulin M (IgM) occurred simultaneously or sequentially [21
]; however, the antibody response to SARS-CoV‑2 and seroprevalence among asymptomatic healthcare workers are far from clear. In this respect the SARS-CoV‑2 immune status of personnel of kidney centers is of eminent importance due to the susceptibility of renal patients to COVID-19.
We aimed to examine the prevalence of SARS-CoV‑2 antibodies among nephrology healthcare workers in a tertiary care, university-based hospital in Austria. We adhered to an orthogonal test strategy and used a comprehensive set of commercial laboratory tests and Western blotting, including COVID-19 controls and analysis of neutralizing antibodies [26
Our study addresses several important issues related to COVID-19: First, the seroprevalence of anti-SARS-CoV‑2 IgG antibodies among nephrology healthcare workers at the Medical University of Vienna during the infection peak in Austria was, at best, 2.1% (95% CI: 0.7–5.0%). Second, it is valid to assume that the successful containment measures taken in Austria and especially in Vienna have minimized exposure of healthcare workers at our institution to COVID-19 in the community and at the point of care as compared to other countries. Third, commercially available laboratory tests, including ELISA, CLIA, and ECLIA, may fail to uniformly detect potential low-level immune response to SARS-CoV‑2 in asymptomatic subjects or mild disease, or differentially cross-react as false positives.
The analytical specificity of a laboratory test is reflected by the positive predictive value (PPV), which depends not only on the sensitivity and specificity of the test but also on the disease prevalence. Requiring a PPV of at least 90%, the analytical specificity of a test should ideally exceed 99.9%, which is influenced by the presence of autoimmune diseases, heterophilic antibodies, or antibodies to other coronaviruses [28
]. At follow-up, we found no effect of influenza vaccination or history of autoimmune disease on COVID-19 serostatus; however, there were more smokers among antibody positives. This finding may be related to the preference of nicotine for the ACE2-SARS-CoV‑2 complex that reduces SARS-CoV‑2 virulence by interfering with the spike protein [30
Overall, serologic tests based on spike glycoprotein appear to distinguish between emerging and endemic coronaviruses, whereas assays based on the nucleocapsid protein can serve as a marker of recent infection but might be expected to cross-react more with endemic coronaviruses [31
]. Test reactivity thresholds used to define a positive result can be adjusted to optimize the trade-off between sensitivity and specificity. With higher thresholds, sensitivity decreases as cases with low serum antibody levels are categorized as negative, but specificity improves as low amounts of nonspecific antibody are no longer considered positive [31
]. In our study, we used a low threshold for IgM and IgG ELISAs at baseline to account for higher sensitivity, accepting low specificity as demonstrated by follow-up examinations.
A meta-analysis of 38 studies covering 7848 individuals confirmed that tests using the spike glycoprotein are more sensitive than nucleocapsid-based tests. The IgG tests performed better compared to IgM tests with higher sensitivity at later time points after the onset of symptoms. Combined IgG and IgM tests performed better in terms of sensitivity than measuring either antibody alone. All methods yielded high specificity with some tests reaching levels around 99% [32
]; however, statistically the PPV varies widely and can be as low as 30–50% in low prevalence settings [33
Rigorous containment strategies may also reduce the prevalence of anti-SARS-CoV‑2 in different populations at the cost of an early development of herd immunity. Travel restrictions and other control measures reduced COVID-19 transmission early last year in China [34
]. Later, it was estimated that among 11 European countries, the national lockdown had the greatest effect on the reproduction number Rt
among nonpharmacologic interventions including school closure, avoidance of public events, social distancing, and self-isolation. As such, Austria (Supplementary Fig. S6) and Norway had the lowest infection rate in this analysis [17
]. At the beginning of this study the Rt
in Vienna was 2.0 (95% CI: 1.87–2.14) and further decreased thereafter. The implementation of public interventions affects the case number after about 2 weeks, and employment of econometric techniques showed that policy changes in 6 countries across the globe averted 530 million infections [35
The low seroprevalence of COVID-19 antibodies in nephrology healthcare workers in our institution reflects successful measures taken to prevent transmission/infection by the city of Vienna and the Medical University of Vienna. Taken together, this minimized exposure risk to COVID-19 for our staff at work (Supplementary Figs. S7 and S8).
The prevalence of asymptomatic cases among SARS-Cov2 infected patients is assumed to be 40–45% [37
]. In our cohort one of five participants considered to be IgG positive showed symptoms potentially related to SARS-CoV‑2 exposure. In contrast, significant exposure to COVID-19 cases resulted in a seroprevalence of 17.4% and 44% among healthcare workers in the USA and in China, respectively [38
]. Other studies in high-risk settings, however, showed a low seroprevalence among hospital staff [40
]. These surveys utilized only one test system and none of these studies employed an orthogonal strategy, confirming borderline positive or positive samples in an independent follow-up serum sample using other laboratory tests.
The extended orthogonal test strategy of the present study, including 12 different commercial tests, Western blots, and a neutralization test, has potentially allowed for an increase in sensitivity/specificity for confirmation of seropositivity among some individuals. Assuming true seropositivity in 5 of 18 healthcare workers, with positive IgG titers in at least 2 of the commercial or in-house test systems at follow-up (uniform nucleocapsid protein IgG in all five cases, also pointing to potential cross-reaction), suggests that single antibody tests do not enable correct detection of true seroconversion and have no acceptable sensitivity and/or specificity in largely asymptomatic and SARS-CoV‑2 RT-PCR negative individuals. This is nicely shown by the rag-rug pattern of seroconversion in Fig. 2
. Of note, nonexposed healthy subjects harbor pre-existing SARS-CoV‑2 cross-reactive T cells, specific for a huge array of SARS-CoV‑2 antigens, suggesting some potential for pre-existing immunity in the population [45
]. This finding matches, at least in part, with the presence of SARS-CoV‑2 antibodies in nonexposed individuals.
In contrast, all 5 COVID-19 serum samples in our study showed SARS-CoV‑2 IgM and IgG antibodies in more than 2 test systems (Fig. 2
). Of the five samples four had neutralizing antibodies, whereas none of the serum samples of the IgG positive study participants showed SARS-CoV‑2 neutralizing capacity (Fig. 2
; Supplementary Table S3). The differential development of antibody response to nucleocapsid protein and spike glycoprotein antigens among patients with proven COVID-19 and the study participants is in support of more cross-reaction than anti-SARS-CoV‑2 seroconversion in healthcare workers enrolled in this study.
A potential limitation to this study lies in the lack of unambiguous COVID-19 cases among study participants, which was not expected to be the case at the beginning of this study in March 2020. We also did not do a formal laboratory test performance analysis. This is largely counterbalanced by the strength of this study, namely the orthogonal test strategy with confirmation in separate serum samples using a very broad range of SARS-CoV‑2 antibody tests.
In summary, our study demonstrates that single antibody tests are not reliable to assess the SARS-CoV‑2 immune response in mostly asymptomatic individuals. This finding has important implications for testing cohorts with a low COVID-19 prevalence to determine whether herd immunity has been reached. Containment strategies by the City of Vienna and the Medical University of Vienna proved to be extremely effective given the very low seroprevalence in a cohort of high-risk healthcare workers during the peak of the pandemic crisis in Austria; however, caution has still to be taken, since healthcare workers are prone to COVID-19 infections and transmission to patients.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.