Demonstrating Clinical Performance of the Elecsys Anti-SARS-CoV-2 Anti-Nucleocapsid Immunoassay Using Real-World Data to Support Regulatory Decision-Making in the USA

Abstract

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) remains a global public health concern. Anti-nucleocapsid (anti-N) serology is a key tool for identifying prior infection and supporting population-level surveillance. This study evaluated the clinical performance of the Elecsys^® Anti-SARS-CoV-2 Anti-N immunoassay on the Cobas^® e 601/602 immunoassay analyzer using real-world data from patients with polymerase chain reaction (PCR)-confirmed symptomatic SARS-CoV-2 infection. The study aimed to provide data which, along with clinical study data, supported the transition from Emergency Use Authorization of the assay to full clearance in the USA.

Methods

We conducted a retrospective review of electronic medical records and laboratory information system data from patients who presented to Columbia University Irving Medical Center from March 2020 to March 2021. Eligible participants were symptomatic, unvaccinated individuals with PCR-confirmed SARS-CoV-2 infection who subsequently underwent serologic testing. Serologic results were categorized by days post symptom onset (DPSO): 0–7, 8–14, and ≥ 15 days. To avoid repeated measures, only the first serologic result per patient within each DPSO category was included. The primary endpoint was positive percent agreement (PPA) for nonimmunocompromised individuals tested at ≥ 15 DPSO, with a prespecified acceptance threshold of ≥ 90%.

Results

A total of 585 serologic tests from 567 patients (77, 45, and 463 samples, in DPSO categories of 0–7, 8–14, and ≥ 15 days, respectively) were evaluated. Among nonimmunocompromised individuals with samples collected ≥ 15 DPSO (N = 285), the PPA was 96.49% (95% confidence interval 93.66%, 98.08%), fulfilling the predefined acceptance criterion. PPA was also consistent across demographic subgroups, supporting the generalizability of the findings.

Conclusions

This study demonstrates the feasibility of generating robust clinical evidence for regulatory purposes using real-world data. The clinical performance of the Elecsys Anti-SARS-CoV-2 immunoassay was confirmed using data collected under routine conditions during the pandemic, complementing data from controlled clinical studies and contributing to informed regulatory decision-making.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40121-026-01336-1.

Key Summary Points

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) remains a global public health concern. Anti-nucleocapsid (anti-N) serology is a key tool for identifying prior infection and supporting population-level surveillance.

This study evaluated the clinical performance of the Elecsys® Anti-SARS-CoV-2 immunoassay on the Cobas® e 601/602 immunoassay analyzer to support its transition from Emergency Use Authorization of the assay to full clearance in the USA.

The study compared retrospective serologic data from patients presenting with polymerase chain reaction (PCR)-confirmed SARS-CoV-2 infection during the pandemic, to assess positive percent agreement (PPA). The prespecified acceptance threshold was ≥ 90%.

The PPA was 96.49% [95% confidence interval 93.66%, 98.08%], fulfilling the predefined acceptance criterion. Consistency across demographic subgroups supports the generalizability of the findings.

This study established the feasibility of generating robust clinical evidence for regulatory purposes using real-world data and further confirmed the clinical performance of the Elecsys Anti-SARS-CoV-2 immunoassay.

Introduction

While the World Health Organization (WHO) declared in May 2023 that the coronavirus disease 2019 (COVID-19) pandemic was no longer a “public health emergency of international concern”, [1] the virus still remains a major public health threat globally. In the USA, nearly 1.2 million people have died as a result of COVID-19 as of June 2024 [2]. Experts have suggested that COVID-19 will be a part of our lives for the foreseeable future [3].

During the pandemic, many new tests for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) were brought to the US market under the Emergency Use Authorization (EUA) process. One example is the Elecsys^® Anti-SARS-CoV-2 immunoassay, intended for qualitative detection of antibodies to SARS-CoV-2 in human serum and plasma [4]. Although serology tests are not intended to diagnose or exclude acute SARS-CoV-2 infection [5], anti-nucleocapsid (anti-N) serology testing can help identify people who may have been naturally exposed to the SARS-CoV-2 virus and have recovered from a SARS-CoV-2 infection. Anti-spike (anti-S) serology testing can also identify those that have been successfully vaccinated and so can be used for monitoring of vaccination effectiveness [6] or estimating the risk of reinfection [7]. Thus, serology tests have played an important role in the control of the COVID-19 pandemic, especially from a population surveillance perspective [8]. With the end of the US Public Health Emergency in May 2023, in vitro diagnostics (IVD) developers must pursue a premarket submission pathway with the Food and Drug Administration (FDA), e.g., de novo, 510(k), or premarket approval (PMA), to ensure the continuous availability of the test on the market. To support such regulatory submissions, there is a need for high-quality evidence generated in the US market that demonstrates the test’s clinical performance. Since 2020, several studies have reported high clinical performance of the Elecsys Anti-SARS-CoV-2 immunoassay across multiple geographic areas, including the USA [4, 9–12]; however, they were not designed for regulatory purposes. A new study, designed specifically to meet the latest FDA requirements in terms of data elements, study design and analytical methods, etc., was needed to support the transition of the Elecsys Anti-SARS-CoV-2 immunoassay from EUA to full clearance in the USA.

Traditionally, the clinical performance of an IVD is measured in a clinical study under a controlled setting. However, in recent years, there have been emerging examples of using real-world data (RWD) or real-world evidence (RWE) to support regulatory decision-making for IVD products [13]. There was a unique opportunity to leverage RWD for the Elecsys Anti-SARS-CoV-2 immunoassay, because during the EUA period, many patients were tested with this assay, yielding a wealth of RWD. Leveraging the RWD to complement data from the clinical study could reduce the burden of sample collection and measurement in the clinical study and accelerate the evidence generation process. As the real-world study population is a subset of the “intended-use population” of this assay, certain subgroups that are typically underrepresented in clinical studies, such as pregnant women or pediatric subjects, may be better represented by the RWD. To facilitate the discussion on using RWD/RWE to transition EUAs for SARS-CoV-2 serology IVDs to 510(k) or de novo applications, the Medical Device Innovation Consortium (MDIC) Open Hand Initiative (a public–private partnership with MDIC, other industry sponsors, and the FDA) was created to allow manufacturers to engage with the FDA in an interactive review process [14].

The objective of this study was to assess the clinical performance of the Elecsys Anti-SARS-CoV-2 immunoassay on the Cobas^® e 601/602 immunoassay analyzers (Roche Diagnostics International Ltd, Rotkreuz, Switzerland) utilizing RWD, among patients with high-sensitivity polymerase chain reaction (PCR)-confirmed, symptomatic SARS-CoV-2 infection. We aimed to use these data in conjunction with data from a clinical study to support the assay’s transition from EUA to full clearance in the USA.

Methods

Ethical Approvals

To ensure adherence to ethical and regulatory standards, the protocol for this study was reviewed and approved by the Columbia University Medical Center Institutional Review Board (IRB-AAAT6416).

Data Source

We leveraged RWD from individuals admitted to or with an encounter at Columbia University Irving Medical Center or its community hospital affiliate, the Allen Pavilion (hereafter referred to as “Columbia University”) from March 2020 to March 2021. All the testing records on the Elecsys Anti-SARS-CoV-2 immunoassay and corresponding high-sensitivity EUA SARS-CoV-2 reverse transcriptase (RT)–PCR during the study period were collected by Columbia University during routine clinical/laboratory practice following the respective laboratories’ internal standard operating procedures. These data were curated directly from the laboratory information system (LIS). All blood samples for the Elecsys Anti-SARS-CoV-2 immunoassay were serum, and the testing was performed on the Cobas e 601 or 602 analyzer.

In addition, for patients’ demographic information (age, gender, race, and ethnicity) and clinical information (e.g., symptom onset date and immunocompromised status), two trained clinical research coordinators (CRCs) independently performed medical chart reviews to extract predefined data elements from the electronic medical record (EMR) system according to a prespecified manual of operations. Specifically, these two professional CRCs each independently extracted the data for the entire cohort following the four-eye principle; each of them took screenshots showing the original source underlying the extracted data element. A Medical Director (MD) reviewed and compared the extracted data from both extractors in real-time. If a discrepancy was observed, the MD first reviewed the screenshots to determine if this was due to data entry errors and made corrections accordingly. If the discrepancy was not due to data entry errors, the MD then reviewed the full source data to resolve the discrepancy and updated the data. In the event of the MD differing from both CRC extractors (i.e., a third answer), the MD’s recommendation was used as the final result. In addition, the MD performed source data verification on a random 10% of all subjects (i.e., reviewing the source EMR to verify if the originally extracted results are correct); if the MD did not agree with the original extraction, they provided the final recommendation, which was used in the analysis. This data extraction process is shown in Fig. 1.

Fig. 1.

Overview of data extraction process of clinical variables. Polymerase chain reaction and serology results were curated directly from the laboratory information system. CRC clinical research coordinator

Study Population

The following inclusion and exclusion criteria for the RWD study were applied.

To be eligible for inclusion, individuals must have presented with COVID-19 symptoms and tested positive for SARS-CoV-2 infection by an EUA authorized, high-sensitivity SARS-CoV-2 PCR test, with sample collection date of the PCR test within 7 days prior to and up to and including 14 days after the date of symptom onset. The date of symptom onset was considered the index date. The date of the blood draw for the Elecsys Anti-SARS-CoV-2 immunoassay serology testing must have been on or after the index date. Individuals were excluded if they had potentially received the COVID-19 vaccination, if they had no date of symptom onset and/or record of symptoms within the required time window around the PCR positive date, if they had the Elecsys Anti-SARS-CoV-2 immunoassay performed before the index date, or if their data had been used in training the data extractors. Samples were classified into three time bins based on DPSO (0–7, 8–14, and ≥ 15 days). Subjects could contribute multiple samples in the real-world setting, so only the first sample in each time bin was included. Remaining (excessive) samples were excluded from analysis.

Definition of Derived Clinical Variables

The onset of COVID-19 symptoms was identified in relation to the first positive PCR test for SARS-CoV-2, based on the primary complaints during the patient’s visit at the respective clinics; the list of symptoms was derived from recommendations made by US Centers for Disease Control and Prevention (CDC) and WHO and in consultation with clinical experts [15, 16]. Symptom definitions are provided in the manual of operations (Supplementary Materials). Immunocompromised status was defined as having one or more of the following conditions that potentially suppress the subject’s immune response: transplantation (e.g., solid organ transplant), comorbidities (e.g., acquired immune deficiency syndrome [AIDS], malignancy except that of skin), or treatment with potential immunosuppressive effects not for treating COVID-19.

Pregnancy status at the time of blood draw for serology testing was determined from the EMR using lab testing and review of notes. Potentially vaccinated individuals were defined as those who received the first dose of COVID-19 vaccination on or before the sample collection date of the first positive PCR, or those with a sample collection date of the first positive PCR on or after 14 December 2020 (when the COVID-19 vaccination commenced in the USA) if the vaccination information was not available.

Statistical Analysis

Statistical analysis was performed on the basis of a prespecified statistical analysis plan. Positive percent agreement (PPA) between the qualitative results from the Elecsys Anti-SARS-CoV-2 immunoassay and the reported PCR-positive SARS-CoV-2 status was calculated as (number of positive serology test results ÷ total number of evaluable samples) × 100%. Calculations were performed using the statistical programming language R Version 4.4.1 (2024) [17]. Two-sided 95% confidence intervals (CIs) for PPA were determined using the Wilson score method [18]. Considering clinical relevance, the primary endpoint was defined as the PPA in nonimmunocompromised subjects using the first serology sample collected ≥ 15 days after symptom onset, as it takes time to develop seroconversion (i.e., transition from infection to the production of antibodies), and this process may be limited among immunocompromised patients, which could bias the assessment of test performance. The acceptance criterion for the primary endpoint was set at ≥ 90.00% for the lower bound of two-sided 95% CI of PPA. Sample size estimation based on formulae proposed previously [19] resulted in a required sample size of 239 assuming a true sensitivity of 95% (power ≥ 0.8 and significance level 0.05). The PPA calculation was further stratified by demographic characteristics, including age, gender, race, and ethnicity.

RWD Quality Assessment

To validate the quality of the manual data extraction from EMRs, a discrepancy analysis was conducted to compare the results between the data extracted by the two CRCs (labeled “E1 versus E2”), and between the 10% data before and after the MD’s verification (labeled “D10% versus R”). The E1 versus E2 comparison captures the differences between the two CRCs, whereas the D10% versus R comparison captures the scenario when the CRCs provided concordant results, but the MD did not agree with this result during the 10% verification round. Figure 1 illustrates the different datasets that were used in the comparisons. The discrepancies were analyzed among the initial cohort, minus a subset that was used for training of extractors for E1 versus E2, and among a group of 10% of subjects selected randomly from the full cohort, less the training subjects for D10% versus R. For each key variable that would directly impact the PPA calculation or help characterize the population, the overall percent agreement and kappa statistics were calculated.

Results

Using an all-comers approach, an initial cohort was identified of 1178 subjects who tested positive on a high-sensitivity EUA SARS-CoV-2 RT–PCR test and were also tested with the Elecsys Anti-SARS-CoV-2 immunoassay during the study period. A total of 611 patients were excluded for the following reasons: 426 patients had no date of symptom onset and/or record of symptoms within the required time window around the PCR-positive date; 139 patients had Elecsys Anti-SARS-CoV-2 immunoassay testing performed before the index date; 41 patients had potentially been vaccinated; and 5 patients’ data were used in training the data extractors. A cohort of 567 subjects was therefore included in the final analysis.

Cohort Characteristics

Among the 567 subjects included in the entire cohort, the median age was 51.0 years, and 48 (8.5%) were pediatric subjects (< 22 years old); 317 (55.9%) were female/250 (44.1%) were male, 155 (27.3%) reported their race as white, and 266 (46.9%) reported their ethnicity as Hispanic or Latino. 95 (16.8%) were assessed as having critical infection.

Among the 366 nonimmunocompromised subjects (i.e., the primary population of interest for PPA calculation), the median age was 44.0 years, and 30 (8.2%) were pediatric subjects (< 22 years old); 234 (63.9%) were female/132 (36.1%) were male, 104 (28.4%) reported their race as white, and 169 (46.2%) reported their ethnicity as Hispanic or Latino. 48 (13.1%) were assessed as having critical infection.

Sample Availability

Within the final cohort of 567 eligible subjects, 782 samples were available. After excluding the samples that were tested before symptom onset, 747 samples were potentially evaluable. After removal of 162 excessive samples and sorting into three time bins, 585 samples from 567 subjects were evaluable resulting in 77, 45, and 463 samples within time bins of 0–7, 8–14, and ≥ 15 DPSO, respectively. The number of samples exceeds the total number of subjects as some subjects contributed to multiple time bins.

Clinical Performance of Elecsys Anti-SARS-CoV-2 Immunoassay

Within nonimmunocompromised subjects, there were 376 evaluable samples, of which 285 samples were drawn ≥ 15 days after the index date; 275 of these 285 samples were positive on the Elecsys Anti-SARS-CoV-2 immunoassay. As shown in Table 1, the PPA was 96.49% (95% CI: 93.66%, 98.08%) within this time bin, which met the acceptance criterion. For completeness, the PPAs within the two earlier time bins are also provided in Table 1.

Table 1.

Positive percent agreement in the nonimmunocompromised population by time bin

Days post symptom onset	Number of samples testeda (with positive RT–PCR test result)	Estimated clinical performance of Elecsys Anti-SARS-CoV-2 immunoassay from RWD
		Reactive (N)	Positive percent agreement (%)	95% confidence interval (%)
0–7	64	34	53.13	41.07, 64.82
8–14	27	18	66.67	47.82, 81.36
≥ 15	285	275	96.49	93.66, 98.08

RT–PCR reverse transcriptase polymerase chain reaction, RWD real-world data

^aSubjects could contribute multiple testing records for the performance calculation; if so, only the first testing record per subject within each time bin was used, and excessive testing records were excluded

To evaluate the generalizability of results, PPA in the nonimmunocompromised population for the third time bin (≥ 15 days) was further stratified by key demographic characteristics including age, gender, race, and ethnicity. The PPA was consistent across subgroups, as shown in Table 2.

Table 2.

Positive percent agreement for nonimmunocompromised subjects using the first sample ≥ 15 days post symptom onset, stratified by gender, age, race, and ethnicity

Group	Number of samples testeda (with positive RT–PCR test result)	Estimated clinical performance of Elecsys Anti-SARS-CoV-2 immunoassay from RWD
		Reactive (N)	Positive percent agreement (%)	95% confidence interval (%)
Gender
Female	181	172	95.03	90.82, 97.36
Male	104	103	99.04	94.75, 99.83
Age (years)
0–21 (pediatrics)	13	12	92.31	66.69, 98.63
22–40	118	112	94.92	89.35, 97.65
41–60	94	93	98.94	94.22, 99.81
61–80	50	49	98.00	89.50, 99.65
81 +	10	9	90.00	59.58, 98.21
Race
Asian	14	14	100.00	78.47, 100.00
Black or African American	50	48	96.00	86.54, 98.90
White	77	73	94.81	87.39, 97.96
Other	94	92	97.87	92.57, 99.41
No information available	50	48	96.00	86.54, 98.90
Ethnicity
Hispanic or Latino	119	115	96.64	91.68, 98.69
Not Hispanic or Latino	116	112	96.55	91.47, 98.65
No information available	50	48	96.00	86.54, 98.90

N/A not applicable, RT–PCR reverse transcriptase polymerase chain reaction, RWD real-world data

^aSubjects could contribute multiple testing records for the performance calculation; if so, only the first testing record per subject within each time bin was used and excessive testing records were excluded

Quantitative Serologic Results from the Full Subject Cohort

Quantitative anti-nucleocapsid serologic results for subjects without evidence of immunosuppression and those who were immunosuppressed are presented in Fig. 2. Although these data were not included in the FDA regulatory submission, analysis of nonimmunosuppressed subjects who underwent serologic testing more than 1 week prior to symptom onset showed that 111 of 114 tested negative, corresponding to a test specificity of 97.4% (95% CI 92.5–99.5). Among the three subjects with positive serology before symptom onset, two had potential EMR evidence of a prior SARS-CoV-2 infection, which was not associated with a documented definitive, positive PCR result. Furthermore, most negative results were well below the manufacturer’s cutoff value of 1.0 cutoff index (COI; signal sample/cutoff). In immunosuppressed subjects, post-symptom onset results often demonstrated an increase in serologic signal that did not reach the positivity threshold, suggesting that a lower cutoff may improve sensitivity for detecting infection in this population. It should be noted that the Elecsys Anti-SARS-CoV-2 assay was not cleared with a quantitative claim, and in the current study, quantitation was only used for research purposes.

Fig. 2.

Quantitative serologic results from the full subject cohort. Quantitative result from the Elecsys Anti-SARS-CoV-2 immunoassay is shown plotted against the time from date of symptom onset to serologic testing in apparently immunocompetent (A) and immunosuppressed (B) subjects. For subjects with multiple serologic tests, only the first test is represented in the figure. Horizontal dotted line represents instrument cutoff for positive serologic test result

Evaluation of Data Quality in Manual EMR Extraction

As described above, we evaluated the discrepancies between data extracted by the two CRCs (E1 versus E2), and between the 10% data before and after the MD’s verification (D10% versus R). Table 3 shows an overview of discrepancies between E1 and E2, and between D10% and R.

Table 3.

Overview of discrepancy for each key variable

Variable	Agreement (%) E1 versus E2a	κ E1 versus E2	Agreement (%) D10% versus Rb	κ D10% versus R
Age	99.9	1.00	100.0	1.00
Gender	100.0	1.00	100.0	1.00
Race	100.0	1.00	100.0	1.00
Ethnicity	99.8	1.00	100.0	1.00
Presence of symptoms	95.9	0.91	100.0	1.00
Date of symptom onset	87.0	N/Ac	98.3	N/Ac
Pregnancy status	93.4	0.88	100.0	1.00
Immunocompromised status	92.5	0.88	100.0	1.00
COVID-19 vaccination status	96.2	0.88	100.0	1.00
Date of first vaccination	98.4	N/Ac	100.0	N/Ac
Time-bin (sample)	94.9	0.91	100.0	1.00

N/A not applicable

^aE1 versus E2 refers to the comparison between the two clinical research coordinators (N = 1173)

^bD10% versus R refers to the comparison within the 10% verification set (N = 118), before and after verification by the Medical Director

^cKappa statistics for dates were not calculated

Discussion

In this study, we demonstrated the clinical performance of the Elecsys Anti-SARS-CoV-2 immunoassay based on RWD from a large academic hospital in the USA. For nonimmunocompromised subjects whose serology samples were collected ≥ 15 days after their symptom onset date for COVID-19 (i.e., the primary population of interest), the PPA met the prespecified acceptance criterion. Results were consistent across different subgroups, supporting the generalizability of these findings to the assay’s intended use population. Being able to leverage RWD generated during the pandemic enabled the design of a smaller clinical study, with evidence generated by each data source complementing the other, leading to the successful transition from EUA to FDA 510(k) clearance for the Elecsys Anti-SARS-CoV-2 immunoassay.

Multiple studies have assessed the performance of the Elecsys Anti-SARS-CoV-2 immunoassay across the USA, Europe, Asia, and Australia, and most have reported high sensitivity/PPA among patients with PCR-confirmed SARS-CoV-2 infection [4, 20–29]. For example, in 2020, a multicenter performance evaluation was conducted at six European sites; the reported sensitivity using samples drawn ≥ 14 days post-PCR confirmation (N = 219) was 93.61% (95% CI 89.51%, 96.46%) [20]. Another study in the UK, led by the National SARS-CoV-2 Serology Assay Evaluation Group, reported a sensitivity of 97.2% (95% CI 95.4%, 98.4%) for this assay using samples collected 20 DPSO [21]. However, most of these studies were performed in a traditional clinical research setting (i.e., the antibody test was performed by the study team). In the scientific community, there have been emerging efforts using RWD to assess the performance of COVID-19 serology assays [30]. RWE can demonstrate the test’s performance in the real-world, intended-use population in the hands of end users under routine conditions, accounting for variability in patient demographics, instruments, control materials, etc. To our knowledge, this study is the first RWD study designed to assess the performance of COVID-19 serology assays according to regulatory evidence requirements. Findings from our study are consistent with clinical studies, which demonstrate the potential for RWD to supplement clinical data in regulatory decision-making.

On the basis of FDA feedback regarding the quality of EMR manual data extraction, we designed a rigorous data extraction process, including initial double extraction under the four-eye principle by two qualified CRCs, as well as additional adjudication and source data verification by the MD. As shown in the results section, the agreements between the two CRCs were above 90% (and often close to 100%) for most variables, the kappa statistics were high, and no discrepancies were reported during the MD’s 10% verification round. The only exception is date of symptom onset, which has an 87% agreement; however, this variable will only affect the analysis through the time bin assignment, which has a 94.9% agreement between the two CRCs. Overall, the low level of discrepancy demonstrates the high quality of the data extraction. More importantly, the MD adjudicated all the discrepancies in the entire cohort and fully verified an additional random 10% data selection, and the MD’s recommendations were used in the final data analysis, thus providing additional assurance for the accuracy of the data used in the final PPA analysis.

This study represents a successful example of generating rigorous clinical evidence for regulatory purposes using RWD. The study protocol, statistical analysis plan and key data elements were predefined through constant engagement and discussion with the FDA through the MDIC initiative. The collaborating institution was carefully selected after comprehensive data landscaping efforts to ensure availability of data elements and prior experience in regulatory studies. Of note, some detailed test data (e.g., test manufacturer and control materials) that would not typically be available from commercial RWD vendors were obtained from the LIS at the collaboration institution, thereby providing comprehensive technical information for the assessment of test performance. In addition, key clinical variables were manually extracted by two well-trained independent extractors and verified by an MD, and discrepancies between different extractors were minimal, demonstrating the high quality of the data extraction. By leveraging existing data curated during routine clinical practice, this well-powered study provided supplementary evidence with consistent conclusions to the traditional clinical study, representing a novel and efficient approach in evidence generation. Furthermore, this is the first diagnostic RWE study registered to the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) RWE Registry, allowing full transparency of the study. Finally, although these analyses were not included in the FDA regulatory submission, further exploration of the RWD demonstrated how such data can be leveraged to assess not only test sensitivity but also specificity. Examining cases in which subjects underwent serologic testing prior to a verified SARS-CoV-2 infection, when they would have been presumed naïve, enabled the estimation of test specificity (currently, however, after multiple waves of SARS-CoV-2 variants, anti-N seropositivity is far more common than previously, which could influence the estimation of test sensitivity). Moreover, quantitative RWD may be able to inform population-specific cutoff values, as several immunosuppressed subjects exhibited antibody responses below the manufacturer’s cutoff yet above the maximum observed in naïve individuals tested before infection. Although our study was not designed to define new thresholds, these observations raise the possibility that population-specific cutoffs may improve clinical interpretation in immunosuppressed populations, in whom antibody responses differ substantially from the general population. Any adjustment to cutoffs would require prospective validation with well-characterized cohorts and correlation to clinical outcomes; however, our RWD potentially supports preliminary evidence that such refinements could be clinically meaningful.

One potential limitation of this study was that COVID-19 vaccination information may have been incomplete, and this information was required as part of the exclusion criteria. However, given the study period (March 2020–March 2021), the majority of the population was identified before vaccination commenced in the USA (December 2020). For those with missing vaccination information, we took a conservative approach and excluded all subjects whose PCR date was after vaccination started in the USA, providing further reassurance that the study population is restricted to those who did not receive COVID-19 vaccination. Moreover, the Elecsys Anti-SARS-CoV immunoassay targets the nucleocapsid protein while only the spike protein is contained in mRNA COVID-19 vaccines.

Another consideration given that the study period precedes widespread COVID-19 vaccine availability (and that during the review process, the FDA stipulated the exclusion of vaccinated individuals) is that the study does not address assay performance in vaccinated or hybrid-immunity populations. An additional limitation of the analysis was that a substantial number of PCR-positive individuals were excluded because symptom onset dates or symptom documentation were unavailable; it is also possible these patients were truly asymptomatic. These exclusions may have introduced selection bias if patients with missing symptom information differ systematically from those with complete records. As a result, the study population may overrepresent individuals with more clearly documented symptomatic disease, which could limit the generalizability of PPA estimates to populations with incomplete symptom reporting.

Conclusions

This study demonstrated the clinical performance of the Elecsys Anti-SARS-CoV-2 immunoassay using RWD collected within routine practice during the pandemic. With similar study design and analytical methods, RWD achieved similar conclusions to the clinical study, and the Elecsys Anti-SARS-CoV-2 immunoassay was successfully transitioned from EUA to full 510(k) clearance. This accomplishment highlights the value of RWD in shaping regulatory decisions and providing clinicians with reliable diagnostic products.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

Eldad A. Hod (conceptualization; investigation; writing—original draft; writing—review and editing). Baiyu Yang (conceptualization; methodology; writing—original draft; writing—review and editing). Carsten Magnus (methodology; formal analysis; writing—original draft; writing—review and editing). Christopher M. Rank (methodology; writing—review and editing). Fei Yang (conceptualization; methodology; writing—review and editing). Annie Qiu (investigation; writing—review and editing). Joseph Lipschitz (investigation; writing—review and editing). Yona Feit (investigation; writing—review and editing). Alex J. Rai (investigation; writing—review and editing). Elodie Baumfeld Andre (conceptualization; methodology; writing—original draft; writing—review and editing).

Funding

This study and participation by Columbia University Medical Center were funded by Roche Molecular Systems Inc. The study sponsor is funding the journal’s publication fee.

Data Availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflicts of Interest

Eldad A. Hod, Joseph Lipschitz, Alex J. Rai, Annie Qiu, and Yona Feit have no conflicts to disclose. Baiyu Yang was an employee of Roche Molecular Systems at the time of this study and holds stock in F. Hoffmann-La Roche AG. Christopher M. Rank is an employee of Roche Diagnostics GmbH and holds stock in F. Hoffmann-La Roche Ltd. Carsten Magnus is an employee of and holds stock in F. Hoffmann-La Roche Ltd. Elodie Baumfeld Andre was an employee of Roche Molecular Systems at the time of this study and holds stocks in F. Hoffmann-La Roche AG. Fei Yang was an employee at F. Hoffmann-La Roche Ltd when the work was conducted. He owns stock at F. Hoffmann-La Roche Ltd. He currently works at Philip Morris International.

Ethical Approval

To ensure adherence to ethical and regulatory standards, the protocol for this study was reviewed and approved by the Columbia University Medical Center Institutional Review Board (IRB-AAAT6416). COBAS and ELECSYS are trademarks of Roche. Any other product names and trademarks are the property of their respective owners. The Elecsys Anti-SARS-CoV-2 immunoassay is approved in the EU for serology/antibody testing. It has an EUA under Section 564 of the US Federal Food, Drug, and Cosmetic Act.US and 510(k) clearance (K250768).