Serological evidence of human infection with SARS-CoV-2: a systematic review and meta-analysis

Chen, Xinhua; Chen, Zhiyuan; Azman, Andrew S; Deng, Xiaowei; Chen, Xinghui; Lu, Wanying; Zhao, Zeyao; Yang, Juan; Viboud, Cecile; Ajelli, Marco; Leung, Daniel T; Yu, Hongjie

BackgroundA rapidly increasing number of serological surveys for anti-SARS-CoV-2 antibodies have been reported worldwide. A synthesis of this large corpus of data is needed. PurposeTo evaluate the quality of serological studies and provide a global picture of seroprevalence across demographic and occupational groups, and to provide guidance for conducting better serosurveys. Data sourcesWe searched PubMed, Embase, Web of Science, and 4 pre-print servers for English-language papers published from December 1, 2019 to September 25, 2020. Study selectionSerological studies evaluating SARS-CoV-2 seroprevalence in humans. Data extractionTwo investigators independently extracted data from studies. Data SynthesisMost of 230 serological studies, representing tests in >1,400,000 individuals, identified were of low quality based on a standardized study quality scale. In the 51 studies of higher quality, high-risk healthcare workers had higher seroprevalence of 17.1% (95% CI: 9.9-24.4%), compared to low-risk healthcare workers and general population of 5.4% (0.7-10.1%) and 5.3% (4.2-6.4%). Seroprevalence varied hugely across WHO regions, with lowest seroprevalence of general population in Western Pacific region (1.7%, 0.0-5.0%). Generally, the young (<20 years) and the old ([≥]65 years) were less likely to be seropositive compared to middle-aged (20-64 years) populations. Seroprevalence correlated with clinical COVID-19 reports, with pooled average of 7.7 (range: 2.0 to 23.1) serologically-detected-infections per confirmed COVID-19 case. LimitationsSome heterogeneity cannot be well explained quantitatively. ConclusionsThe overall quality of seroprevalence studies examined was low. The relatively low seroprevalence among general populations suggest that in most settings, antibody-mediated herd immunity is far from being reached. Given the relatively narrow range of estimates of the ratio of serologically-detected infections to confirmed cases across different locales, reported case counts may help provide insights into the true proportion of the population infected. Primary Funding sourceNational Science Fund for Distinguished Young Scholars (PROSPERO: CRD42020198253).