Diagnostic Efficacy of 11 SARS-CoV-2 Serological Assays for COVID-19: A Meta-Analysis and Adjusted Indirect Comparison of Diagnostic Test Accuracy

Abstract

Objective: In the past 5 years, a large number of serological assays for large-scale detection of antibodies against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antigen emerged. Serological assays for SARS-CoV-2 were needed to support clinical diagnosis and epidemiological investigations. However, there were limited data on the diagnostic accuracy of these serological assays. We aimed to compare the diagnostic accuracy of 11 commercial serological assays for coronavirus disease-2019 (COVID-19) by taking the reverse transcriptase polymerase chain reaction (RT-PCR) assays as the reference standard, which served as the control arm to conduct an indirect comparison of diagnostic accuracy for 11 different SARS-CoV-2 serological assays.

Methods: This meta-analysis was conducted following the PRISMA 2020 reporting guideline. Electronic searches were performed using the Cochrane Library, PubMed, Embase, Web of Science, Chinese Biological Medicine Database (CBM), China National Knowledge Infrastructure (CNKI), WANFANG, and Chinese Weipu (VIP) databases. Fifty-seven articles, including 11 serologic-based IgG, IgM, and total antibodies assays for SARS-CoV-2, published before June 2024, were included in this meta-analysis. The main outcome of this meta-analysis used to evaluate the performance of 11 assays included pooled diagnostic odds ratio (DOR), area under the summary receiver operating characteristic (AUC), and summary receiver operating characteristic curve (SROC). The R software was used for adjusted indirect comparison to calculate the relative diagnostic odds ratio (RDOR) with corresponding 95% confidence intervals (CIs), and indirect comparison forest plots showed the results.

Results: A total of 57 articles met the eligibility criteria for inclusion in our meta-analysis. The pooled DOR and the AUC for access SARS-CoV-2 IgG were 564.28 (95% CI 229.58-1386.91) and 1.00, and as for EDI novel coronavirus COVID-19 IgG those were 85.27 (95% CI 53.99-134.68) and 0.95, for EDI novel coronavirus COVID-19 IgM were 49.42 (95% CI 16.47-148.30) and 0.86, for iFlash-SARS-CoV-2 IgG were 652.31 (95% CI 362.32-1174.41) and 0.97, for iFlash-SARS-CoV-2 IgM were 36.72 (95% CI 12.42-108.54) and 0.76, for MAGLUMI 2019-nCoV IgG were 145.44 (95% CI 59.37-356.30) and 0.90, for MAGLUMI 2019-nCoV IgM were 21.59 (95% CI 14.27-32.67) and 0.59, for ortho-clinical anti-SARS-CoV-2 IgG were 719.46 (95% CI 262.34-1973.13) and 1.00, for ortho-clinical anti-SARS-CoV-2 total were 1104.60 (95% CI 395.64-3083.99) and 1.00, for Siemens SARS-CoV-2 total (COV2T) were 1143.37 (95% CI 316.49-4130.62) and 0.99, for Wantai SARS-CoV-2 total Ab were 1014.98 (95% CI 618.48-1665.66) and 1.00. The pooled DOR for assays-based IgG (n = 43), assays-based total antibody (n = 35), and assays-based IgM (n = 20) was 242.88 (95% CI 157.66-374.16), 1215.90 (95% CI 547.14-2702.07), and 40.99 (95% CI 22.63-74.25). The diagnostic accuracy of assays-based total antibody performed better than those of assays-based IgG and assays-based IgM; assays-based IgG performed better than assays-based IgM.

Conclusion: This study suggested that the Siemens SARS-CoV-2 total (COV2T), ortho-clinical anti-SARS-CoV-2 total, and Wantai SARS-CoV-2 total had the best overall diagnostic accuracy. The diagnostic efficacy of the assays-based total antibody had statistically significantly higher accuracy than those of assays-based IgG and assays-based IgM for COVID-19.

Keywords: COVID‐19; SARS‐CoV2; adjusted indirect comparison; meta‐analysis; serological assays.

Figure 1

Flow diagram of selecting the literature and screening process.

Figure 2

Risk of bias and applicability concerns summary.

Figure 3

(A) Pooled DOR with 95% CIs of immunoglobulin classes. (B) Indirect comparison forest plots of RDOR with 95% CIs for all three pairwise immunoglobulin comparisons. CIs, confidence intervals; DOR, diagnostic odds ratio; RDOR, relative diagnostic odds ratio.

Figure 4

Forest plots of coupled sensitivity and specificity with 95% confidence intervals for 11 serological assays. Harritshøj (A): Siemens Vista assay was performed, Harritshøj (B): Siemens Atellica assay was performed; Ward (A): Siemens Atellica assay was used, Ward (B): Siemens EXL systems were used; Zilla (A): Siemens Centaur assay was performed; Zilla (B): Siemens Vista assay was performed.

Figure 5

(A) SROC curve of the 11 serological assays. (B) Pooled DOR with 95% CIs of the 11 serological assays. (C) Indirect comparison forest plots of RDOR with 95% CIs for three IgM assays pairwise comparisons. CIs, confidence intervals; DOR, diagnostic odds ratio; RDOR, relative diagnostic odds ratio; SROC, summary receiver operating characteristics.

Figure 6

Indirect comparison forest plots of RDOR with 95% CIs for IgG and total antibody assays pairwise comparisons. CIs, confidence intervals; RDOR, relative diagnostic odds ratio.