Evaluation of 32 rapid tests for detection of antibodies against SARS-CoV-2

Tollånes, Mette C.; Jenum, Pål A.; Kierkegaard, Helene; Abildsnes, Eirik; Magne Bævre-Jensen, Roar; Breivik, Anne C.; Sandberg, Sverre

Aims To evaluate the analytical performance of 32 rapid tests for detection of antibodies against coronavirus SARS-CoV-2. Materials and methods We used at total of 262 serum samples (197 pre-pandemic and 65 convalescent COVID-19), and three criteria to evaluate the rapid tests under standardized and optimal conditions: (i) Immunoglobulin G (IgG) specificity “good” if lower limit of the 95% confidence interval was ≥97.0%, “acceptable” if point estimate was ≥97.0%, otherwise “not acceptable”. (ii) IgG sensitivity “good” if point estimate was ≥90.0%, “acceptable” if ≥85.0%, otherwise “not acceptable”. (iii) User-friendliness “not acceptable” if complicated to perform or difficult to read result, otherwise “good”. We also included partial evaluations of three automated immunoassay systems. Results Sensitivity and specificity varied considerably; IgG specificity between 90.9% (85.9-94.2) and 100% (97.7-100.0), and IgG sensitivity between 53.8% (41.9-65.4) and 98.5% (91.0-100.0). Combining our evaluation criteria, none of the 28 rapid tests that detected IgG had an overall performance considered “good”, seven tests were considered “acceptable”, while 21 tests were considered “not acceptable”. Four tests detected only total antibodies and were not given an overall evaluation. IgG sensitivity and/or specificity of the automated immunoassays did not exceed that of many rapid tests. Conclusion When prevalence is low, the most important analytical property is a test’s IgG specificity, which must be high to minimize false positive results. Out of 32 rapid tests, none had a performance classified as “good”, but seven were classified as “acceptable”.