RESUMO
INTRODUCTION: Serological surveys offer the most direct measurement to define the immunity status for numerous infectious diseases, such as COVID-19, and can provide valuable insights into understanding transmission patterns. This study describes seroprevalence changes over time in the context of the Democratic Republic of Congo, where COVID-19 case presentation was apparently largely oligo- or asymptomatic, and vaccination coverage remained extremely low. METHODS: A cohort of 635 health care workers (HCW) from 5 health zones of Kinshasa and 670 of their household members was interviewed and sampled in 6 rounds between July 2020 and January 2022. At each round, information on risk exposure and a blood sample were collected. Serology was defined as positive when binding antibodies against SARS-CoV-2 spike and nucleocapsid proteins were simultaneously present. RESULTS: The SARS-CoV-2 antibody seroprevalence was high at baseline, 17.3% (95% CI 14.4-20.6) and 7.8% (95% CI 5.5-10.8) for HCW and household members, respectively, and fluctuated over time, between 9% and 62.1%. Seropositivity was heterogeneously distributed over the health zones (p < 0.001), ranging from 12.5% (95% CI 6.6-20.8) in N'djili to 33.7% (95% CI 24.6-43.8) in Bandalungwa at baseline for HCW. Seropositivity was associated with increasing rounds adjusted Odds Ratio (aOR) 1.75 (95% CI 1.66-1.85), with increasing age aOR 1.11 (95% CI 1.02-1.20), being a female aOR 1.35 (95% CI 1.10-1.66) and being a HCW aOR 2.38 (95% CI 1.80-3.14). There was no evidence that HCW brought the COVID-19 infection back home, with an aOR of 0.64 (95% CI 0.46-0.91) of seropositivity risk among household members in subsequent surveys. There was seroreversion and seroconversion over time, and HCW had a lower risk of seroreverting than household members (aOR 0.60 (95% CI 0.42-0.86)). CONCLUSION: SARS-CoV-2 IgG antibody levels were high and dynamic over time in this African setting with low clinical case rates. The absence of association with health profession or general risk behaviors and with HCW positivity in subsequent rounds in HH members, shows the importance of the time-dependent, and not work-related, force of infection. Cohort seroprevalence estimates in a 'new disease' epidemic seem insufficient to guide policy makers for defining control strategies.