A total of 597 households were selected from 105 health areas (clusters), of which 585 (97.9%) agreed to participate in the survey. From the selected households, 2,681 persons were surveyed, of whom 2,601 (97%) were present on the survey dates. From the 2,601 eligible persons, only 2,560 (98.4%) consented to be interviewed and provided blood samples. Of the 2,560 blood samples, 58 dried blood spots were excluded. We successfully processed 2,502 (97.7%) samples, including 26 (1%) samples with borderline results that were not included in the analysis (Figure 2).

Of the 2,560 eligible participants, 1,412 (55.2%) were female and 1,148 (44.8%) were male. Most participants (1,787, 69.8%) were from health districts located in the western part of Kinshasa (Table 1). The median age was 30 years (interquartile range 18–46 years). Persons 20–29 years of age were the most represented (511/2,560, 19.9%), followed by those 30–39 years of age (409, 15.9%). The median size of the household was 7 (interquartile range of 6–9). Six households in 10 (349/585, 59.6%) reported an average monthly income of $51–$250 (US dollars), whereas nearly one quarter (145/585) reported an average monthly income of $1–$50. Most participants (1,209/2,560, 47.2%) had junior-high school education level, and 3.0% (77) had no formal education (Table 1).

Regarding COVID-19 prevention measures, 37.9% (971/2,560) of participants reported washing their hands >6 times/day; another one fifth reported washing their hands daily (481 reported 1×/d, 535 reported 2×/d). More than one quarter of participants reported wearing a face mask frequently (659, 25.7%), but nearly one quarter reported rarely wearing a face mask (623, 24.3%). Most participants were nonsmokers (2,410, 94.1%) and 67.2% (1,721) reported no alcohol consumption (Table 1).

One in 2 participants reported >1 symptom indicative of COVID-19; fever was mentioned most frequently (713, 27.8%), followed by headache (627, 24.5%), chills (423, 16.5%), fatigue (409, 15.9%), and cough (400, 15.6%), (Table 2). Only 12.0% (308) of participants reported >1 comorbidity, with hypertension (166, 6.5%) and obesity (96, 3.7%) being the most reported. Most participants reported no contact with a laboratory-confirmed COVID-19 case (2,493, 97.4%) (Table 2).

The overall population-weighted SARS-CoV-2 seroprevalence was 72.2% (95% CI 69.8–4.4%) (Appendix Table 2). The seroprevalence was slightly higher, although not significantly, among female than male participants (73.8% vs. 70.1%; p = 0.146), and significantly higher in the western health districts of Kinshasa than in the eastern (74.3% vs. 68.3%; p = 0.021) (Appendix Table 1). Two health districts on either side of Kinshasa had the highest seroprevalence: Barumbu (88.4%, 95% CI 74.9%–95.1%) and Masina 2 (88.6%, 95% CI 77.1%–94.8%) (Appendix Table 2). Similarly, higher seroprevalence was found among participants 40–49 years of age (78.6%, 95% CI 72.9%–83.3%), those with university education (84.0%, 95% CI 69.9%–92.2%), and those who declared washing hands 5–6 times a day (76.8%, 95% CI 71.9%–81.0%) (Appendix Table 1). Most households (94.2%, 551/585) had >1 seropositive member; median was 3 positive members (Appendix Table 3, Figures 2, 3). After adjusting for the laboratory test kit performance, the overall seroprevalence increased from 72.2% (95% CI 69.8%–74.4%) to 76.5% (95% CI 74.5%–78.5%).

Participants living in households with an average monthly income of $51–$250 had 42% increased odds of SARS-CoV-2 infection compared with participants from a household with an average monthly income of $1–$50 (crude OR 1.42, 95% CI 1.12–1.80) (Table 3). In contrast, participants from households with an average monthly income of >$1,000 were 88% less likely to be infected with SARS-CoV-2 (crude OR 0.12, 95% CI 0.04–0.36). The likelihood of SARS-CoV-2 infection tended to increase with increasing education level and age. Participants with university-level education were >3 times more likely to be infected with SARS-CoV-2 than those without formal education (crude OR 3.73, 95% CI 1.10–12.67) (Table 3).

On multivariable analysis, after adjusting for sex, age, and geographic area, an association emerged between SARS-CoV-2 infection and the 5–9-year age group (adjusted OR 0.38, 95% CI 0.15–0.96) and the 10–14-year age group (adjusted OR 0.33, 95% CI 0.12–0.91) (Table 3). SARS-CoV-2 infection remained associated with average household monthly income, especially for a household earning >$1,000 (adjusted OR 0.12, 95% CI 0.04–0.33). The association with education level remained and became stronger in effect size and statistical significance, especially for participants with university-level education (adjusted OR 4.33, 95% CI 2.36–17.18) (Table 3).

We conducted this population-based serologic survey at the end of the second wave (October 2020–April 2021) of COVID-19 in Kinshasa, DRC, before vaccination became available. As such, our results provide evidence of the cumulative exposure to SARS-CoV-2 among this population.

Our results show that, 1 year after detecting the first COVID-19 case, >3 out of 4 persons (76.5%) had been infected with SARS-CoV-2. This high seroprevalence indicates sustained community transmission in Kinshasa. Considering the Kinshasa population of 15 million in March 2021 (15), we estimate 11.5 million SARS-CoV-2 infections had occurred by March 6, 2021, but only 19,831 confirmed cases were reported (1 PCR-confirmed case for nearly 580 estimated infections). During the same period, 3,325 COVID-19 cases were active, and <100 hospitalizations occurred in COVID-19 treatment centers. Factors such as the younger age of the population, the predominance of mild and asymptomatic cases self-managed in the community, poor testing capacities, and low healthcare utilization might explain the discrepancies between reported cases, the actual number of infections, and the number of moderately or severely ill hospitalized persons.

The population-weighted and test kit–adjusted seroprevalence of 76.5% was nearly 4 times higher than that reported after the first wave (16.6%), reflecting an extensive community transmission after the lightening of lockdown measures, including the travel ban implemented during the first wave (5). Lower seroprevalence estimates were reported in Bangladesh (63.1%), Mali (58.5%), India (54.2%), Zimbabwe (53.0%), Kenya (44.2%), and Sierra Leone (2.6%) during similar periods (16–22). The characteristics of the test kit used and the variability in exposure levels across countries might explain differences in seroprevalence estimates. In our study and that of Bangladesh, an ELISA-based test detecting total antibodies against the RBD of the spike protein was used (10,16). Studies from Zimbabwe and India used serologic assays, targeting IgGs directed against the nucleocapsid protein, which are known to wane faster over time than those directed against the spike protein (18,19,23). Studies from Mali and Kenya used serologic tests that only targeted anti-spike IgG, thus missing newly infected persons who could bear anti-spike IgM rather than IgG (17,20). The Sierra Leone study used a lateral flow assay that targets both anti-spike IgM and IgG but is less sensitive than ELISA (22). Female participants were nearly 20% more likely to be infected than male participants, but the difference was not statistically significant. Similar results have been reported from other sub-Saharan Africa countries (24,25). As reported after the first wave of the COVID-19 epidemic in the DRC, the seroprevalence was not statistically different between western and eastern health districts of Kinshasa on multivariable analysis, although more cases were reported in western health districts (5). In our study, we observed a trend of increasing seroprevalence with age. This trend is consistent with other reports from Africa and Asia, which found higher exposures among participants 39–59 years of age (16,17,20,22). In addition, our results suggest that the second wave was characterized by similar infection rates for all age groups (5).

The risk for SARS-CoV-2 infection increased with average monthly household income up to $500 before decreasing dramatically, especially among households with incomes of >$1,000. Household income was associated with SARS-CoV-2 infection, and higher incomes reduced the risk for infection in households (26). The discrepancy in our study can be explained by respondent bias, because household income was assessed based on household heads’ responses rather than owned assets. Participants with a university education were more likely to be infected (84%), and having a university-level education was associated with a 4-fold increase in the risk for infection. A study from Portugal reported that a lower level of education was a critical risk factor for SARS-CoV-2 transmission compared with tertiary education (27). Higher levels of education are usually associated with better job opportunities and higher income and, thus, better living conditions and compliance with individual and collective protective measures. Conversely, a higher level of education and better employment may be associated with higher mobility and complex interactions with potentially infected persons, increasing the odds of infection.

Our study has several strengths, such as the robust sampling frame, which provided a large and representative sample size that included all 35 health districts of Kinshasa, and the high response rate among households (97.9%) and participants (98.4%). However, we were unable to perform a neutralizing antibody test on positive samples to ascertain protective immunity. In addition, there might have been respondent bias because we relied on self-reporting for variables such as comorbidities, household income, face mask wearing, daily hand washing, alcohol intake, and tobacco use. The stigma associated with COVID-19 might have played a role in underreporting critical information, as exemplified by the lower proportion (2.6%) of participants who reported a known contact with a laboratory-confirmed case. Finally, we collected clinical symptoms by interviewing participants, and there might have been recall bias, especially for symptoms that occurred more than 2 weeks prior to the survey.

In conclusion, our data suggest an extensive transmission of SARS-CoV-2 during the second COVID-19 wave in Kinshasa, resulting in a higher seroprevalence. Evidence generated from this population-based survey is critical to adjusting the COVID-19 response and especially vaccination campaign strategies in the context of vaccine scarcity and hesitancy, when a large proportion of potential vaccinees have been naturally exposed to SARS-CoV-2. The emergence and global spread of SARS-CoV-2 variants of concern, with their potential to resist neutralizing antibodies developed after natural infection, and antibodies waning could hamper the putative protective immunity. Serosurveillance coupled with neutralization tests and genomic surveillance of SARS-CoV-2 variants is needed to adjust the COVID-19 response plan in the DRC, including vaccination strategies, as the pandemic evolves.