Introduction

Children are less likely than adults to be infected and develop severe disease by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)1,2. Throughout the COVID-19 pandemic, serological assays have been essential in guiding pandemic responses3. Serological assays have multiple clinical applications, such as analyzing the immune response, detecting asymptomatic cases, and identifying individuals with potential immunity4.

Various SARS-CoV-2 binding antibody assays have been developed, targeting different antigens and utilizing various assay formats5. Serological assays are particularly effective in the later stages of infection when the RT-PCR results may be inconclusive6. Specific antibodies such as IgM, IgA, and IgG serve as diagnostic markers during this phase, appearing within 5–15 days of infection. While IgM and IgA antibodies generally remain detectable for 3–6 weeks, IgG antibodies can persist for several months7. Therefore, IgG antibodies can be valuable for the diagnosis of past infections.

Most serological assays target the receptor-binding domain (RBD) of the spike (S) protein, S protein, nucleocapsid (N) protein, or a combination of these antigens due to their high expression and immunogenicity8,9. Serological assays use various methodologies, including Enzyme-Linked Immunosorbent Assay (ELISA), Chemiluminescent Immunoassay (CLIA), and lateral flow immunoassay (LFIA)10. LFIAs are quick but less sensitive, making them ideal for rapid testing11. ELISA and CLIA offer higher sensitivity and the ability to test multiple samples simultaneously. Although CLIA provides more precise quantification, it is also more expensive compared to ELISA12,13.

While data on the early immune response and seroconversion are limited, the available binding antibody assays vary significantly, despite ongoing research on their clinical relevance and performance5,14,15. Despite increasing use in SARS-CoV-2 seroprevalence research, qualitative antibodies can be challenging to interpret and apply for clinical and public health decision-making16,17,18. Due to the varying levels of sensitivity and specificity among assays, interpreting serology results requires caution. In this study, we used multiple detection methods to compare SARS-CoV-2 antibody test results from a sample of children across 60 schools in the Oromia Region. The three assays compared included: in-house ELISA anit-SARS-CoV-2 RBD, developed by Armauer Hansen Research (AHRI) and Oregon Health and Sciences University (OHSU), hereafter called in-house ELISA19, the Elecsys® anti-SARS-CoV-2 anti-nucleocapsid assay (Roche Diagnostics, Basel, Switzerland)20, and a pan-Ig Rapid LFA test kit (Hangzhou Covid-19 IgG/IgM rapid test device)21. The in-house ELISA detects antibodies against the RBD of the SARS-CoV-2 spike protein, specifically targeting IgG in serum or plasma specimens. This assay has a reported sensitivity of 100% when analyzing serum samples collected two weeks after symptom onset, with a specificity of 97.7% in pre-pandemic samples19. In contrast, the Elecsys CLIA anti-SARS-CoV-2 assay, which detects anti-nucleocapsid antibodies using the Cobas 6000 e601 system, has a specificity of 99.80% and a sensitivity of 99.5% in samples collected more than 14 days after PCR confirmation22. The Rapid LFA test (IgM & IgG) demonstrated a sensitivity of 96.7% and a specificity of 93.7%11. Antibody tests possess the potential to elucidate prior SARS-CoV-2 infections within populations, and may subsequently facilitate the assessment of immune responses to vaccination.

Furthermore, antibody assays can contribute to the clinical diagnosis of COVID-19 patients in instances where qRT-PCR yields negative results despite a strong clinical suspicion. The present study entails a comparative evaluation of an in-house ELISA to determine its utility for SARS-CoV-2 testing. It used samples collected before the omicron pandemic, in parallel with two commercially available antibody assays that utilize different assay formats and directed at several antigens.

Methods

Study design and settings

A longitudinal, school-based cohort study was conducted between December 2020 and April 2021 in hotspot zones and towns of the Oromia Region in Ethiopia before the emergence of the omicron variant23. The Armauer Hansen Research Institute (AHRI), Additionally, engages in disease and genomic surveillance, and policy studies, and provides training across various research and development programs in collaboration with both domestic and international research institutes.

Human sera (n = 1441) for this study were collected in December 2021 from students at 60 randomly selected schools in the top 15 hotspot areas within the Oromia Region, encompassing eight zones and seven town administrations.

Data collection and laboratory procedures

In April 2021, a standardized questionnaire was used to collect demographic data, information on COVID-19-related symptoms, and details about prevention practices. To ensure accuracy and efficiency, the data collectors, sample collectors, and supervisors underwent two days of training: one day focused on the survey tools and another on the laboratory components. The questionnaires were pre-tested before implementation, and investigators supervised the supervisors and data collectors throughout the process. The collected data were promptly sent online to the central server at Oromia Health Bureau for storage. Additionally, GPS coordinates were recorded for all schools.

Blood specimen collection and laboratory antibody testing

Specimen collection, transportation and storage

Approximately 5 ml venous blood was collected in a serum separator tube (SST) following standard laboratory practices and the manufacturer’s instructions. The collected serum was stored at − 20 °C or below until it was ready for use, with care taken to avoid repeated freezing and thawing cycles. When shipping the specimens, they were packaged and labelled correctly per local and international regulations for transporting clinical specimens and pathogenic agents. The in-house ELISA and Rapid LFA Test (IgM and IgG) were performed on similar freeze–thaw serum samples in duplicates. Due to the unavailability of the reader for Elecsys CLIA at the Adam Regional Laboratory, frozen aliquots were shipped to Jimma University. Consequently, the testing of freshly thawed serum samples in duplicates using CLIA was conducted at Jimma University.

Serological methods

To evaluate the concordance of the in-house ELISA developed at AHRI, Addis Ababa, Ethiopia with two commercial serological assays: the Hangzhou Covid-19 IgG/IgM rapid test device (Hangzhou Biotest Biotech Co., Ltd., Hangzhou, China); and the Elecsys CLIA (Roche Diagnostics, Basel, Switzerland), all 1,441 serum samples were tested for the presence of anti-SARS-CoV-2 antibodies using these three assays in duplicates in April 2021. The commercial serological assays were performed according to manufacturers’ protocols, while the in-house ELISA testing followed previously described procedures19. Testing of the serum samples using in-house ELISA and Rapid LFA Test (IgM and IgG) were performed at the Adma Regeion Laboratory in similar freeze–thaw serum samples in duplicates. However, due to the unavailability of the reader for Elecsys CLIA at the Adam Regional Laboratory, frozen aliquots were shipped to Jimma University. Consequently, the testing of freshly thawed serum samples in duplicates using CLIA was conducted at Jimma University.

In-house AHRI ELISA

The in-house indirect ELISA is semi-quantitative and measures serum immunoglobulin G (IgG) against the RBD of the SARS-CoV-2 spike protein. Optimization and validation of this assay were performed using pre-pandemic sera collected between 2012 and 2018 and plasma or serum samples collected from confirmed COVID-19 at different time points post-onset of symptoms19. The present serum samples were tested following the previously described in-house indirect ELISA procedures19. Briefly, the microtiter plates were coated with purified recombinant proteins of receptor binding domain of the spike protein of SARS-CoV-2 (100 μl/well) diluted in phosphate-buffered saline, PBS (pH 7.4) at concentration 1 μg/mL and incubated overnight at 4 °C. Next day, excess unbound antigen was removed and thereafter microtiter plates were blocked with 300 μl/well of 4% skimmed milk with PBS plus 0.1% Teween-20 (w/v) for 2 h at room temperature (RT). Following the blocking step, microtiter plates were washed 3X with PBS plus 0.05% Tween-20 (PBST) and thereafter 100 μl/ml of serum sample diluted at 1:200 in blocking buffer was added and incubated at RT for 60 min. Following incubation and 5X washes with PBST, 100 μl/well of horseradish peroxidase-conjugated anti-human IgG (Invitrogen, USA) diluted at 1:5000 in blocking buffer was added and incubated for 1 h at RT. After 5X washes, the reaction was visualized by adding 75 μl/well 3,3′,5,5′-Tetramethylbenzidine (TMB) liquid substrate (Bio-Rad, USA) and incubating at RT in the dark for 10–15 min. The reaction was then stopped with 75 μl/well TMB stop solution. The optical density (OD) was measured at 450 nm filter on an ELISA LT-45000 microplate reader. Each sample was tested in duplicate.

Quality control

To ensure quality control, AHRI provides hands-on training for all laboratory personnel involved in serum sample testing using in-house ELISA at the study site. The training covers procedures and protocols for in-house ELISA testing, including the preparation of coating, blocking, and washing buffers, the interpretation of test results, and ongoing competency assessments. AHRI also supplies the necessary materials for training, processing, and testing serum samples using in-house ELISA. These materials include SARS-CoV-2 RBD antigen-coated ELISA plates, secondary antibodies, substrate solutions, and true-positive and true-negative serum samples for anti-SARS-CoV-2 antibodies to be included in each in-house ELISA run as internal quality control. Two professors from Jimma University supervised the serology tests. Sample results underwent a double data entry process by two data managers, and no discrepancies were found.

Data management and analysis

The data was exported from CSentry CSPro DataEntry 7.2.1 to Microsoft Excel for cleaning and then transferred to Stata version 14.2 for analysis. We calculated various diagnostic parameters such as categorial variables as absolute numbers and percentages and Cohen’s kappa coefficient (κ). Kappa statistics were interpreted as follows: less than 0.2 represents poor agreement; 0.2–0.4 represents fair agreement; 0.41–0.6 represents moderate agreement; 0.61–0.8 represents substantial agreement; and greater than 0.8 represents a great agreement24. We also assessed the Total Percent Agreement (TPA), Positive Percent Agreement (PPA), and Negative Percent Agreement (NPA) between the three serology assays. The diagnostic parameters were presented with a 95% Confidence Interval (CI) determined using the Clopper-Pearson method.

Ethical considerations

Ethical approval for the study was obtained from the Oromia Health Bureau Review Board and Jimma University Institute of Health (Ref # BEFO/HBTFH/1-16/234 and JUIH/IRB/333/21). Support letters from the Oromia Education Bureau were also secured for selected schools. Written informed consent was obtained from parents and/or legal guardians. All information collected during the study was kept confidential through anonymity. Participants suspected of having COVID-19 were referred to the nearest health facility for appropriate care. All methods were performed in accordance with the Ethiopian Public Health Institute guidelines and regulations.

Results

SARS-COV-2 seroprevalence varied by location, age, and sex, with results differing between assays

The serological investigation was carried out in 1441 serum samples collected from fifteen districts and towns that reported the highest prevalence of SARS-CoV-2 cases within the specified period. The highest positivity rate was observed in Gasera District, with a 74.3% positivity rate determined by the Rapid LFA test (IgG & IgM). There is a statistically significant association among various zones and towns in relation to sero prevalence (P Value = 0.001) while in-house ELISA was employed. Dukem and Sebata towns also showed positivity rates exceeding 50% across all SARS-CoV-2 antibody assays (Fig. 1).

Fig. 1
figure 1

Positivity rate by location from serum samples collected before the emergence of the Omicron variant in Ethiopia. (N = 1141).

Full size image

The mean age of the participants was 15.8 years (SD: 2.6). Among all age groups, the highest positivity rate was observed in individuals over 18 years of age regardless of the testing method used, ranging from 49.9% for in-house ELISA to 58.7% for Rapid LFA test (IgM & IgG.) However, there was no statistically significant difference in positivity rates between different age groups (P value = 0.409) compared with the in-house ELISA. Similarly, there was no significant association between having underlying health conditions (P value = 0.154) or contact with confirmed cases and antibody positivity (P value = 0.608) and having contact with confirmed cases by in-house ELISA. Differently, there was a statistically significant difference in seropositivity rates between males and females (P value = 0.045) seroprevalence. In-house ELISA revealed that the overall seroprevalence, or the proportion of people in a community who had contracted SARS-CoV-2 at a given point in time in the past, was 42.7% (95% CI 39.8–45.6) (Table 1).

Table 1 Comparison of the performance of SARS-CoV-2 antibody assays and positivity rate across various scenarios.
Full size table

Kappa agreements of in-house ELISA with two commercially available SARS-CoV-2 serologic assays

The overall percentage agreement between in-house ELISA and Elecsys CLIA was 80.8% (95% CI 75–86.5), indicating strong agreement. The percentage agreement with the Rapid LFA test (IgG + IgM) was slightly lower, at 75.8% (95% CI 70.1–81.5) (Table 2). The kappa coefficients (κ), which measure agreement beyond chance, were: κ = 0.61 (95% CI 0.55–0.67) for in-house ELISA and Elecsys CLIA, indicating substantial agreement; κ = 0.52 (95% CI 0.46–0.58) for in-house ELISA and the Rapid LFA test (IgG + IgM), indicating moderate agreement; and κ = 0.73 (95% CI 0.67–0.78) for Elecsys CLIA and the Rapid LFA test (IgG + IgM), indicating substantial agreement. Additionally, the concordance rate between the rapid LFA test (IgG only) and Elecsys CLIA was 89.9% (95% CI 83.2–94.8), with a κ of 0.79 (95% CI 0.73–0.85), indicating substantial agreement (Table 2).

Table 2 Total agreements and Kappa coefficient among In-house ELISA and two commercially available serological assays.
Full size table

Positive and negative percent agreement

The positive percent agreement (PPA) and negative percent agreement (NPA) between the in-house ELISA and the two commercial assays are summarized in Table 3. The in-house ELISA demonstrated strong agreement with the Elecsys CLIA, showing a PPA of 81.7% and an NPA of 80.1%. Compared to the Rapid LFA test (IgG + IgM), which had a PPA of 83% and an NPA of 70.4%, the in-house ELISA exhibited better overall agreement with Elecsys CLIA.

Table 3 Positive and negative percent agreements between in-house ELISA and two commercially available serological assays.
Full size table

Interestingly, although the PPA between the Rapid LFA test (IgG + IgM) and Elecsys CLIA (81.1%) was comparable to that between the in-house ELISA and Elecsys CLIA, the Rapid LFA test (IgG + IgM) displayed a relatively higher NPA of 91.9%. This indicates that while both the in-house ELISA and the Rapid test (IgG + IgM) had similar PPA with Elecsys CLIA, the Rapid LFA test (IgG + IgM) was more consistent in correctly identifying negative samples when compared to Elecsys CLIA.

Discussion

The main goal of this study was to compare the performance of in-house ELISA with two established serological assays for detecting SARS-CoV-2 antibodies, LFA, and CLIA, using large (n = 1441) serum samples crossectionally collected from students in various districts and towns in Oromia region, Ethiopia. These areas were considered hotspots for SARS-CoV-2 transmission based on RT-PCR-positive COVID-19 cases reported before the Omicron pandemic. Our study showed significant variability in SARS-CoV-2 seroprevalence by location, age, and sex of study students who submitted the serum samples. The overall seroprevalence was 42.9% by in-house ELISA, 47.1% by Rapid LFA Test, and 46.3.7% by Elecsys CLIA. Statistically significant higher SARS-CoV-2 seropositivity rates (49.3%, 52.2%, and 58.7%) were found in students over 18 years old compared to students under 18 years old (42.9%, 49.6%, and 54.2%). Higher seropositivity rates were especially noted in males and those with direct contact with confirmed cases, underscoring the impact of exposure on transmission dynamics.

The performances of the three serological assays in this study were compared based on seropositivity, percentage agreement, and level of concordance using the Kappa value scale (Landis and Koch, 1977). The percentage agreement between the Elecsys CLIA and in-house ELISA was relatively high at 80.8%, with substantial kappa agreement (κ = 0.61), indicating significant diagnostic accuracy. In contrast, κ values between the in-house ELISA and the Rapid LFA test (IgG only) was 0.52, highlighting moderate variability in diagnostic accuracy. Although the percentage agreement observed between the Elecsys CLIA and the Rapid LFA test (IgG + IgM) was higher at 86.2% compared to the agreement between in-house ELISA and Elecsys CLIA, the kappa (κ = 0.73) values still demonstrated within the range of substantial agreement (k = 0.61–0.80). This overall agreement supports the performance of the in-house ELISA is comparable to the two commercial serological assays, highlighting its potential as a reliable, cost-effective alternative for large-scale seroepidemiological surveys.

Existing literature supports these findings, indicating that laboratory-based assays like Elecsys CLIA tend to have higher diagnostic accuracy than Rapid LFA test, particularly for detecting IgG antibodies25,26. Similarly, research findings emphasize that assays targeting spike and nucleocapsid proteins may be more effective in identifying past infections. Our results align with these studies, suggesting the utility of IgG-based assays for long-term serological surveys in regions with high prior exposure27,28.

The modest disagreement among the compared serology assays may be attributed to the differences in their target proteins; the In-house ELISA is designed to target the RBD of the SARS-CoV-2 spike protein, whereas the Elecsys targets the nucleocapsid protein of SARS-CoV-2. The assay methods differed: the in-house ELISA is a colorimetric assay, while Elecys is CLIA. Various studies have compared serology assays for SARS-CoV-2, highlighting differences in performance and results. Another study compared four quantitative SARS-CoV-2 serological assays and found poor correlation in quantitative, semi-quantitative, and qualitative measurements, emphasizing the need for assay harmonization29. Additionally, research on SARS-CoV-2 antibody levels post-vaccination revealed dynamic changes across different assay systems, offering valuable insights for standardization efforts30. An external quality assessment among laboratories in EU/EEA countries demonstrated a solid capacity to detect SARS-CoV-2 antibodies; however, performance varied based on the assay method used.

The in-house ELISA exhibits better agreement (k ≥ 0.6) with the Rapid LFA test (IgG only) and Elecys CLIA. This can be due to the similarity of the three serology assays targeting similary antibodies (IgG). Overall, assay methods that detect total immunoglobulins, neutralizing antibodies, or IgG have demonstrated better performance than IgA- or IgM-specific31,32,33,34,35.

Furthermore, the moderate to high agreement between the in-house ELISA and commercial assays supports the feasibility of locally developed assays as cost-effective alternatives, especially in resource-limited settings where imported kits may not always be accessible.

These findings have important public health implications. Serological assays are crucial for understanding the epidemiology of SARS-CoV-2, particularly in areas with high vaccination rates or limited diagnostic capabilities for qRT-PCR. In resource-constrained regions, reliable and cost-effective antibody assays can guide public health measures, including monitoring population immunity and evaluating vaccine effectiveness36. The substantial agreement between the in-house ELISA and Elecsys CLIA highlights the potential utility of the in-house ELISA as a feasible and reliable serological tool for large-scale seroepidemiological surveys.

Additionally, the lack of concordance between assays detecting different proteins (RBD vs. nucleocapsid) underscores the necessity of strategic assay selection based on the intended use, whether for determining past exposure or monitoring vaccine responses.

Our results demonstrated that the in-house ELISA, akin to the Elecsys CLIA, can quantify antibody levels and serves as a high-throughput assay, processing numerous samples within a few hours. Moreover, it can detect individuals who had a SARS-CoV-2 infection without previous molecular diagnostic confirmation, highlighting its potential use as a cost-effective serosurveillance tool to estimate the actual burden of SARS-CoV-2 infection in the country.

Serological surveys such as this one provide valuable insights into true infection rates in the community, especially in asymptomatic or mild cases that may not have been captured by routine diagnostic testing. They can also identify potential gaps in vaccine coverage or waning immunity, thereby informing decisions on booster campaigns and other public health interventions.

Strengths and limitations

The study’s strengths include its substantial sample size and the diverse clinical scenarios of the participants. Additional testers implemented rigorous quality control measures, and data were collected across sixty schools in eight zones and fifteen districts/towns. The survey spanned a wide geographical area, with samples gathered before the onset of the Omicron variant and the commencement of vaccination campaigns in the country. Notably, the in-house ELISA stands out as the only domestically manufactured SARS-CoV-2 testing methodology.

However, the study had limitations, including the absence of data on symptom onset dates, which hindered the assessment of serology assay performance across pre-symptomatic, asymptomatic, and symptomatic cases. Additionally, the lack of RT-PCR confirmation for samples means that the duration of post-exposure at the time of blood sample collection was unknown, limiting the ability to explain the discordance among the three assays fully.

Conclusion

This study highlights significant variability in SARS-CoV-2 antibody positivity across different regions in Ethiopia before the emergence of the Omicron variant. The findings suggest substantial agreement between the in-house ELISA and Elecsys CLIA and a modest agreement between the in-house ELISA and the Rapid LFA Test (IgG + IgM). Despite this, the in-house ELISA demonstrated strong potential as an effective and cost-efficient tool for anti-SARS-CoV-2 antibody testing. It is a valuable option for guiding public health responses and supporting seroepidemiological surveillance studies, particularly in resource-limited settings. Continued monitoring of SARS-CoV-2 exposure and immunity is essential to inform policies better and prepare for future outbreaks.