Skip to main content

Serial interval of novel coronavirus (COVID-19)

Our take —

This study estimated the serial interval of COVID-19 utilizing data from transmission pairs through February 12, 2020. The data for this analysis were from publicly shared data across six settings. The two approaches yielded estimates for the serial interval of 4.0 and 4.6 days, which were shorter or similar to the incubation period. This study highlighted the high potential for pre-symptomatic transmission and potential challenges to contract tracing.

Study design


Study population and setting

Overall, 28 transmission pairs (each comprised of an identified infector and infectee) from 12 family clusters through February 12, 2020, contributed to the dataset compiled from publicly shared transmission pairs from Vietnam, South Korea, Germany, Taiwan, China, and Singapore. The date of illness onset was defined as the date on which a symptom relevant to COVID-19 infection appeared and was determined by the reporting governmental body.

Summary of Main Findings

This study estimated a median serial interval of 4.0 days (95% credible interval [CrI]: 3.1, 4.9), accounting for right truncation and analyzing all pairs. The median serial interval was estimated as 4.6 days (95% CrI: 3.5, 5.9) when limiting the data to only the most certain pairs.

Study Strengths

The analysis adjusted using right truncation to account for the potential bias due to sampling at an early stage of the epidemic which may preferentially exclude transmission pairs with longer serial intervals. The authors also conducted sensitivity analyses limiting data to only those determined most credible by the authors.


The process for selecting transmission pairs was not described. Further, the transmission pairs were assessed for credibility, however this process was only described as a subjective assessment.

Value added

This study compared the estimates of serial intervals from different geographic settings using a dataset of transmission pairs with varying levels of credible data to a limited dataset of transmission pairs determined to be credible. At the time, studied with linked pairs were limited to a very small number of pairs.