Interpretation of SARS-CoV-2 serologic test results, except pan Igs Wanti ELISA, has been reported to be very challenging in Africa due to pre-existing cross-reactive antibodies induced by other pathogens such as non-SARS-CoV-2 human coronaviruses and malaria parasites. Given the rapid decline of anti-SARS-CoV-2 nucleocapsid antibodies as compared to the anti-RBD IgG antibody13, we developed and optimized an in-house ELISA that detects anti-SARS-CoV-2 IgG antibodies. Our assay, unlike other commercially available serologic assays, is affordable and has been validated with a large number of Ethiopian sera from both pre-COVID-19 and COVID-19 patients from the same regions. Its sensitivity on convalescent sera from COVID-19 patients confirmed by RT-PCR was found to be as sensitive as the Wantai pan Ig ELISA (100%), and superior to Realy Tech’s IgM/IgG LFA (90%). Also, our in-house assay displayed 97.7% specificity in randomly selected pre-COVID-19 Ethiopian origin sera, which is superior to Realy Tech (92.5 %).
Seroprevalence studies provide information about the extent of individuals who had exposure to to the virus and help to understand the future course of the pandemic and are key to providing target prevention and control measures in reducing transmission and severe outcomes. In this study, the overall seroprevalence of SARS-CoV-2 spike RBD IgG antibodies among HWs was 39.6%, ranging from 24.5% in the Hiwot Fana Specialized Hospital, Harar to 48·0% in ALERT Hospital located in the capital city, Addis Ababa. This is not a surprise given Addis Ababa is the epicenter of SARS-CoV-2 transmission in Ethiopia, and SARS-CoV-2 has been introduced 4 months later in Harar. As a result of which, it is expected that a higher proportion of HWs in hospitals located in Addis Ababa, including ALERT are frequently exposed to COVID-19 cases than that HWs working in hospitals located in Harar, where fewer number cases and deaths had been reported.
According to our finding, at least 4 in 10 urban Ethiopian HWs had already been exposed to SARS-CoV-2 by February 2021 in Ethiopia. This result contrasts with a serosurvey in asymptomatic individuals from the general population conducted in March 2020 in Addis Ababa (8.8%) and from the household serosurveys in Jimma (2%) and Addis Ababa (5%) that were conducted during the first wave of the pandemic-i.e., four months after the first COVID-19 case in Ethiopia. Although this stark seroprevalence difference between our study and these two previous studies might be explained by differences in the types of assays employed, lack of personal protective equipment (PPE) and/or respective cohorts, the most plausible explanation is that the sera for the present serosurveillance study had been collected after the first wave of the pandemic in Ethiopia, between March 2020 and February 2021.
While the high seroprevalence rates observed among the different geographically located hospitals are approaching those of high-incidence countries like Brazil, they are in agreement with several other SARS-CoV-2 seroprevalence studies from sub-Saharan Africa that, like Ethiopia, have reported much lower rates of RT-PCR confirmed cases and deaths. For example, higher anti-SARS-CoV-2 antibody seroprevalence has been reported in South Sudan (30–60.6%) , Democratic Republic of Congo (8%-36) and Nigeria (25%-45)  depending on the population sampled and the serological test use. Taken together these studies indicate that SARS-CoV-2 has spread widely in sub-Saharan Africa . However, the majority (74.0%) of our study participants never had any symptoms compatible with COVID-19, suggesting the occurrence of significant burden of asymptomatic infections and its transmissions in the country, which is now, being reflected in the trend of increasing PCR positivity since January 2021. The higher proportion of younger HWs (mean age of 34 years), and the fewer participants with comorbidities (6.7%) may have contributed to the observed high burden of asymptomatic infection among the studied HWs. Malaria, BCG-vaccination, warmer environment, and high prevalence of pre-existing cross-reactivate against HCoVs may have also contributed.
A report from Spain showed a higher (38.3%) seroprevalence of SARS-CoV-2 among HWs . This is comparable with the present report from Ethiopia, where there were a relatively fewer severe cases and deaths. Similarly, higher seroprevalence among frontline HWs has been reported in other sub-Saharan African countries such as in Malawi . These findings and ours highlight the importance of asymptomatic infections in the African countries. Interestingly, we found no seroprevalence differences between healthcare occupations including administrative staff. The lack of a dramatic difference between front line HWs and administrators may be a reflection of the frontline administrative staff are also at high risk and are poorly protected, or may suggest the level of virus transmission in the general population at large as previously observed in UK. Nevertheless, further well-designed investigations are required to implement occupation-specific public health strategies in healthcare facilities.
In the present study, a history of previous close contact with a suspected or confirmed COVID-19 case was found to be strongly associated with seropositivity; however, this finding contradicts the observed similar seropositivity between front line HWs and administrators. Similar odds of seropositivity between males and females were also found although several studies elsewhere reported higher odds of seropositivity in males . A similar contradictory finding was reported in the Spanish general population.
Our study has several strengths. These include its use of an in-house developed assay which we optimized to significantly minimize false positive responses by validating it with both pre-pandemic and pandemic samples of Ethiopian origin. Most importantly, the study involved a relatively large sample size from five hospitals located in different geographical locations, providing much needed information about the COVID-19 pandemic in sub-Saharan Africa.
Despite these strengths, our study has several limitations. First, all hospital staff were invited to take part in the study, and hence selection bias might have affected our results. Second, recall bias might have affected the responses to the history of symptoms compatible with COVID-19, and close contact with a confirmed COVID-19 case, and thereby contributed to the absence of a strong correlation between seropositivity and these covariates, albeit having close contact with COVID-19 case. Third, our findings are slightly affected by the accuracy of our assay, with a sensitivity of 100% in convalescent samples from RT-PCR conformed COVID-19 cases and specificity of 97.7% in pre-COVID-19 samples. However, even this slight overestimation of the apparent seroprevalence associated with the assay specificity is likely to be matched by the proportion of study participants who might be infected and yet not produce humoral immune responses at the time of blood sample collection.