Characterization of unlinked cases of COVID-19: Implication on contact tracing measures.

BACKGROUND Contact tracing and intensive testing programmes are essential for COVID-19 control. However, conventional contact tracing is resource intensive and is unable to trace all the cases due to recall bias and anonymity of a number of close contacts. So far, few studies have reported the epidemiological features of cases not identified from contact tracing ("unlinked cases") or described their potential roles in seeding community outbreaks. OBJECTIVE Here we characterized the role of unlinked cases in the epidemic by comparing their epidemiological profile with the linked cases and estimated their transmission potential across different settings. METHODS We obtained the government rapid surveillance data containing the line-listing of COVID-19 confirmed cases accounting for the first three waves in Hong Kong. We compared the demographics, history of chronic illnesses, epidemiological characteristics, clinical characteristics, and outcomes between linked and unlinked cases. Transmission potentials at different settings were assessed by fitting a negative binomial distribution to the observed offspring distribution. RESULTS Time interval from illness onset to hospital admission was longer among unlinked cases than the linked (median: 5.00 vs 3.78 days; P<.001), with a higher proportion experiencing critical or serious conditions (13.0% vs 8.2%; P<.001). The proportion of unlinked cases was associated with the increase of the weekly number of local cases (P=.049). Cluster transmissions from the unlinked cases were most frequently identified in household settings, followed by eateries and workplaces, with the estimated probability of cluster transmissions being around 0.4 for households and 0.1-0.3 for the latter two settings. CONCLUSIONS The unlinked cases were positively associated with the time to hospital admission, severity of infection, and epidemic size - implying a need to design and implement digital tracing methods to complement current conventional testing and tracing. To minimize the risk of cluster transmissions from unlinked cases, digital tracing approaches should be effectively applied in high-risk socio-economic settings, and risk assessment conducted to review and adjust the policies.