Temporal variation in the serial interval of SARS-CoV-2 driven by non-pharmaceutical interventions

Abstract

Background: Studies of novel coronavirus disease (COVID-19) have reported varying estimates of epidemiological parameters including serial interval distributions, i.e. the time between illness onset in successive cases in a transmission chain, and reproduction numbers. While the serial interval is an essential metric for estimating many other key epidemiological parameters including reproduction number, which are in turn used to predict disease trends and healthcare demands. Method: In this study, we examined the possible reason of these diverse estimates of the parameters. We first compiled a database and reconstructed the transmission pairs for COVID-19 in mainland China. We estimated the mean serial intervals for COVID-19 from this pair data using normal distribution fitting on pre, post and during peak weeks. We also proposed the effective serial interval, a measure of real-time serial intervals allowing for variation over time. We verified the effects of non-pharmaceutical interventions on the serial interval over time by using probabilistic models, individual simulation models and regression models. Finally we showed how the estimation of reproduction number can be corrected by using effective serial interval distributions. Results: We reconstructed of 1,407 transmission pairs from a line-list database on COVID-19 in mainland China, in which 677 transmission pairs had information on symptom onset dates and social relationships for both the infector and infectee. By compiling these transmission pairs, we show that mean serial intervals of COVID-19 have shortened substantially from 7.8 days to 2.6 days within a month (January 9 to February 13, 2020). This change is driven by enhanced non-pharmaceutical interventions, in particular case isolation. We also show that using effective serial intervals, it provides more accurate estimates of reproduction numbers than using conventionally fixed serial interval distributions. Conclusion: Our results indicate that caution is needed when attempting to generalize estimates of the serial interval distribution to other places or to other periods in the same place, for example when estimating instantaneous reproductive numbers These findings would improve assessment of transmission dynamics, forecasting future incidence, and estimating the impact of control measures.