RASSEGNA   

La Rivista Italiana della Medicina di Laboratorio 2021 Dicembre;17(4):227-36

DOI: 10.23736/S1825-859X.21.00122-5

Copyright © 2021 EDIZIONI MINERVA MEDICA

lingua: Inglese

Diagnostic accuracy of serological tests for COVID-19: a systematic review and meta-analysis of cohort studies

Clístenes C. DE CARVALHO ^{1, 2} ✉, Moara M. CARDOZO ^{1, 3}, Rebeca GONELLI ¹, Stéphanie L. REGUEIRA ⁴, Ana B. SOUZA ⁴, Isabella B. RAMOS ⁴, Caroline O. REGO ⁴, Marielle B. MANOEL ⁴, Antônio S. NETO ⁴, Kalil L. FERRAZ ⁵, Vanessa O. SANTOS ⁶, Arthur L. AZEVEDO ⁶, Felipe S. CALDEIRA ⁶, Débora C. BELFORT ³, Carlos A. DE ARAÚJO III ³

¹ Department of Post-Graduation, Prof. Fernando Figueira Institute of Integral Medicine (IMIP), Recife, Brazil; ² Department of Surgery, Federal University of Campina Grande, Campina Grande, Brazil; ³ Center of Biological Sciences and Health, Catholic University of Pernambuco, Recife, Brasil; ⁴ Center of Biological Sciences and Health, Federal University of Campina Grande, Campina Grande, Brazil; ⁵ Faculty of Medical Sciences, University of Pernambuco, Recife, Pernambuco, Brazil; ⁶ Faculdade Pernambucana de Saúde (FPS), Recife, Brazil

INTRODUCTION: Although RT-PCR is considered the current reference for COVID-19 diagnosis, it is time and resource demanding as well as it presents some intrinsic inaccuracy. Complementary tools such as serological tests might therefore be of value throughout the diagnostic process. We then aimed to evaluate the diagnostic accuracy of serological tests for ongoing COVID-19.
EVIDENCE ACQUISITION: We conducted a search in PubMed, Embase, Web of Science, LILACS, Scielo, Cochrane COVID-19 Study Register, medRxiv, and bioRxiv in September 2020. We included cohort studies fully reported with patients at all ages, providing data on diagnostic accuracy of serological tests for ongoing COVID-19 - ascertained by RT-PCR. We excluded studies published in language other than English, Spanish, French, or Portuguese and those we were unable to abstract relevant data. We performed article screening, risk of bias judgments and data extraction in duplicate by 2 independent reviewers with ability to resolve conflict with supervising authors. A random-effects meta-analysis using hierarchical models was performed. Sensitivity, subgroup, and meta-regression analyses were also applied. We judged quality of the evidence using GRADE recommendations.
EVIDENCE SYNTHESIS: Of the initial 21,312 screened articles, 10 studies evaluating 2,824 patients were included. All studies were at high risk of bias and low concern regarding applicability. There was significant heterogeneity among the studies, but without threshold effect. Summary sensitivity and specificity were 68.1% (95% CI: 54.6-79.1%) and 91.7% (95% CI: 78.4-97.1%), respectively. The estimated AUC was 0.84. The evidence was regarded to be of very low quality.
CONCLUSIONS: The assessed evidence is not enough to determine diagnostic accuracy of the serological tests for COVID-19. The studies are scarce, heterogeneous, imprecise and of very low quality.

KEY WORDS: COVID-19; SARS-CoV-2; Serological tests; Sensitivity and specificity; Systematic review