We obtained plasma samples from 1,028 volunteers >18 years of age during March 1–June 29, 2022, in the course of the Omicron BA.2 epidemic wave. All participants reported that, before the sampling date, they had never tested positive for COVID-19 by reverse transcription PCR test or RAT. We tested plasma samples using ELISA with ORF8 protein as an antigen for detection (Appendix). We obtained daily numbers of reported cases confirmed by PCR or RAT from the Hong Kong Department of Health. On the basis of the rates of ORF8 ELISA–positive test results relative to the number of reported cases at different time points, we estimated the true daily numbers of SARS-CoV-2 infections and infection attack rates by fitting a multinomial-distribution model, accounting for sensitivity and specificity of RAT and PCR tests. We assumed an initial infection attack rate of 0.2% before 2022, given Hong Kong’s stringent infection control measures before the Omicron outbreak (2,7). We also reconstructed the time-varying reproduction number by renewal equation (8). We used Markov chain Monte Carlo method to estimate the posterior distributions of model parameters, summarized by medians with 95% credible intervals (CrIs) (Appendix).

Of the 1,028 self-reported uninfected persons in our study, 371 (36.1%) were male and 657 (63.9%) female, and median age was 50 (range 18–88) years; 1,027 reported having received >2 doses of COVID-19 vaccines. Overall positivity rate of ORF8 ELISA testing among our cohort was 2.5% (26 positive/1,028 tested). We found positivity rates were unlikely to vary on the basis of sex, age, or calendar date among self-reported uninfected persons in our cohort (Table).

Figure

Figure. Reported SARS-COV-2 incidence versus estimates based on open reading frame 8 antigen testing, Hong Kong, China, January 1–June 20, 2022. A) Daily numbers and incidence of all reported infections and...

Among the total population in Hong Kong, 16.2% were reported to have tested positive by RAT (6.1%) or reverse transcription PCR (10.1%). On the basis of estimates from our statistical model (Appendix), we inferred that 33.6% (95% CrI 32.1%–34.8%) of the 7.5 million persons in Hong Kong were infected during January 1–June 20, 2022 (Figure, panels A, B), corresponding to ≈2.5 million persons. We calculated percentages of asymptomatic cases of 41.8% among reported SARS-CoV-2 infections and 72.1% (95% CrI 70.8%–73.0%) among total (reported and unreported) infections. Reproduction numbers obviously dropped after positive RAT result reporting was implemented in Hong Kong on February 26, 2022 (Figure, panel C), consistent with findings about changes in transmission dynamics reported elsewhere (9).

Using the seroprevalence of ORF8 antigens, we inferred that 16.2% of 33.6% (≈1/2) SARS-CoV-2 infections during the Omicron epidemic in Hong Kong were unrecognized, despite RATs being widely disseminated and reporting of positive results made locally compulsory. Our estimates of asymptomatic proportions were generally higher than estimates previously reported for earlier variants (10). With such a large number of unrecognized cases circulating the virus in the community, it was not surprising that the Omicron outbreak was uncontrollable, even though stringent measures, such as contact tracing and quarantine for close contacts, continued to be in effect. Our study findings highlight the usefulness of testing for ORF8 seroprevalence among efforts to monitor COVID-19 outbreaks, especially for emerging new variants of concern. Public health agencies need to take into account the potential for substantial undercount of actual numbers of infections when considering the commitment of resources to prevent and control outbreaks.