Table 5

Accuracy of covid-19 serology tests stratified by potential sources of heterogeneity*†

Potential source of heterogeneityNo of armsTPFNPooled sensitivity (95% CI)TNFPPooled specificity (95% CI)
Peer reviewed:
 ELISA: Yes1077219083.7 (74.1 to 92.5)19161399.3 (98.7 to 99.7)
 ELISA: No861312985.3 (75.7 to 91.9)14523998.4 (94.6 to 99.7)
 LFIA: Yes744610672.4 (37.9 to 93.6)4391896.4 (88.3 to 99.3)
 LFIA: No32156676770.3 (52.1 to 83.7)293516097.3 (94.9 to 98.6)
 CLIA: Yes82495389.7 (61.3 to 98.7)76910586.5 (80.3 to 99.4)
 CLIA: No9197037390.5 (73.2 to 97.4)43858699.1 (91.8 to 100)
Data level:       
 ELISA: Patient858913388.2 (67.1 to 96.9)17231898.8 (97.0. 99.5)
 ELISA: Sample1179618682.2 (76.4 to 87.1)16553499.8 (97.8 to 100)
 LFIA: Patient1567521170.7 (46.9 to 86.1)7994397.2 (92.8 to 99.4)
 LFIA: Sample24133766275.1 (56.3 to 95.0)257513597.6 (97.6 to 99.4)
 CLIA: Patient11149018590.7 (80.3 to 96.8)391518588.4 (48.8. 98.4)
 CLIA: Sample672924189.3 (45.8 to 96.1)1239699.8 (98.4 to 100)
Antigen target
ELISA:       
 Surface protein857316680.8 (68.1 to 89.9)16003598.3 (94.9 99.6)
 Nucleocapsid protein538910878.3 (72.5 to 83.2)6701598.2 (92.1 to 99.8)
 Surface and nucleocapsid proteins54234592.5 (80.4 to 98.1)1098299.9 (99.8 to 100)
LFIA:       
 Surface protein659419167.4 (35.7 to 89.4)4395192.6 (82.6 to 98.7)
 Nucleocapsid protein12629-450-
 Surface and nucleocapsid proteins736311978.9 (64.4 to 91)9672098.7 (91.9 to 100)
CLIA:       
 Surface protein1355197-3340-
 Nucleocapsid protein1727-6416-
 Surface and nucleocapsid proteins10142015291.6 (72.7 to 98.2)46019298.4 (94.8 to 99.7)
Commercial kit:       
 ELISA: Yes1093722880.9 (73.9 to 86.5)13573598.7 (94.4 to 99.8)
 ELISA: No184489187.3 (72.4 to 95.3)20111799.3 (98.1 to 99.8)
 LFIA: Yes36154781265.0 (49.0 to 78.2)320616597.0 (94.8 to 98.5)
 LFIA: No34656188.2 (83.6 to 91.3)1681394.9 (79.1 to 99.8)
 CLIA: Yes72276184.2 (61.4 to 96.3)4918390.8 (19.8 to 99.7)
 CLIA: No8165129693.9 (75.6 to 98.9)45159999.2 (87.0 to 100)
Population for estimating specificity†
ELISA:       
 Samples collected before covid-19 epidemic7   13603098.7 (93.4 to 99.9)
 Samples collected during covid-19 epidemic in individuals not suspected of having covid-1910-- 1441999.5 (97.9 to 100)
 Individuals with suspected covid-19 but RT-PCR negative result-------
 Individuals with confirmed other viral infection5-- 4252196.3 (62.2 to 100)
LFIA:       
 Samples collected before covid-19 epidemic25   20717998.2 (96.1 to 99.7)
 Samples collected during covid-19 epidemic in individuals not suspected of having covid-197---518599.0 (97.8 to 100)
 Individuals with suspected covid-19 but RT-PCR negative result7---4244390.7 (87.5 to 93.2)
 Individuals with confirmed other viral infection12   4455690.8 (76.2 to 95.7)
CLIA       
 Samples collected before covid-19 epidemic1---64218-
 Samples collected during covid-19 epidemic in individuals not suspected of having covid-197  -416712597.8 (86.7 to100)
 Individuals with suspected covid-19 but RT-PCR negative result8-- 2429277.1 (15.6 to 98.4)
 Individuals with confirmed other viral infection1---3340-

TP=true positive; FN=false negative; TN=true negative; FP=false positive; ELISA=enzyme linked immunosorbent assay; LFIA=lateral flow immunoassay; CLIA=chemiluminescent immunoassay; RT-PCR=reverse transcriptase polymerase chain reaction

  • * Pooled estimates were calculated using bivariate random effects meta-analysis; as such, they depart from what would be estimated by simple division of numerator by denominator. Supplementary table S4 lists the names of commercial assays.

  • Pooled estimates were calculated using univariate random effects meta-analysis.