Insights into COVID-19 epidemiology and control from temporal changes in serial interval distributions in Hong Kong

• Ali, Sheikh Taslim ^[1] ; Chen, Dongxuan ^[1] ; Lau, Yiu-Chung ^[1] ; Lim, Wey Wen ^[1] ; Yeung, Amy ^[1] ; Adam, Dillon C ^[1] ; Lau, Eric H Y ^[1] ; Wong, Jessica Y ^[1] ; Xiao, Jingyi ^[1] ; Ho, Faith ^[1] ; Gao, Huizhi ^[1] ; Wang, Lin ^[2] ; Xu, Xiao-Ke ^[3] ; Du, Zhanwei ^[1] ; Wu, Peng ^[1] ; Leung, Gabriel M ^[1] ; Cowling, Benjamin J ^[1]
  1. [1] University of Hong Kong

     University of Hong Kong

     RAE de Hong Kong (China)

     
  2. [2] University of Cambridge

     University of Cambridge

     Cambridge District, Reino Unido

     
  3. [3] Dalian Minzu University College of Information and Communication Engineering, , Dalian, Liaoning Province, 116600

  Mostrar afiliaciones +
• Localización: American journal of epidemiology, ISSN-e 1476-6256, ISSN 0002-9262, Vol. 194, Nº. 4, 2025, págs. 1079-1089
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Abstract The serial interval (SI) distribution of an epidemic is used to approximate the generation time distribution, an essential parameter for inferring the transmissibility (${R}_t$) of an infectious disease. However, SI distributions may change as an epidemic progresses. We examined detailed contact tracing data on laboratory-confirmed cases of COVID-19 in Hong Kong, China, during the 5 COVID-19 waves from January 2020 to July 2022. We reconstructed the transmission pairs and estimated time-varying effective SI distributions and factors associated with longer or shorter intervals. Finally, we assessed the biases in estimating transmissibility using constant SI distributions. We found clear temporal changes in mean SI estimates within each epidemic wave studied and across waves, with mean SIs ranging from 5.5 days (95% credible interval, 4.4-6.6) to 2.7 days (95% credible interval, 2.2-3.2). The mean SIs shortened or lengthened over time, which was found to be closely associated with the temporal variation in COVID-19 case profiles and public health and social measures and could lead to biases in predicting ${R}_t$. Accounting for the impact of these factors, the time-varying quantification of SI distributions could lead to improved estimation of ${R}_t$, and could provide additional insights into the impact of public health measures on transmission.