Variability in transmission risk of SARS-CoV-2 in close contact settings: A contact tracing study in Shandong Province, China

Abstract

Background: Understanding the relative transmissibility of SARS-CoV-2 virus across different contact settings and the possibility of superspreading events is important for prioritizing disease control. Such assessment requires proper consideration of individual level exposure history, which is made possible by contact tracing.

Methods: The case-ascertained study in Shandong, China including 97 laboratory-confirmed index cases and 3158 close contacts. All close contacts were quarantined after their last exposure of index cases. Contacts were tested for COVID-19 regularly by PCR to identify both symptomatic and asymptomatic infections. We developed a Bayesian transmission model to the contact tracing data to account for different duration of exposure among individuals to transmission risk in different settings, and the heterogeneity of infectivity of cases.

Results: We estimate secondary attack rates (SAR) to be 39% (95% credible interval (CrI): 20-64%) in households, 30% (95% CrI: 11-67%) in healthcare facilities, 23% (95% CrI: 7-51%) at workplaces, and 4% (95% CrI: 1-17%) during air travel. Models allowing heterogeneity of infectivity of cases provided a better goodness-of-fit. We estimated that 64% (95% CrI: 55-72%) of cases did not generate secondary transmissions, and 20% (95% CrI: 15-26%) cases explained 80% of secondary transmissions.

Conclusions: Household, healthcare facilities and workplaces are efficient setting for transmission. Timely identification of potential superspreaders in most transmissible settings remains crucial for containing the pandemic.

Keywords: COVID-19; Close contact; Households; Superspreading; Transmission risk.

Fig. 1

The distribution for number of close contacts per each case overall and by contact type. The curve indicated the expected number from fitted distribution.

Fig. 2

The distribution for number of secondary cases per each primary case. The curve indicated the expected number from fitted distribution.

Fig. 3

Scatter plots of model-predicted vs. observed number of secondary cases in each close contact group, by assumption about the mean incubation period and the maximum infectious period. The predicted numbers are from the models include individual-level random effects. Closeness of dots to the diagonal dashed line indicates agreement between model prediction and observation. Dotted lines serve as error band for a difference of ≤ 1 between model-predicted and observed numbers.