Serial Intervals and Case Isolation Delays for Coronavirus Disease 2019: A Systematic Review and Meta-Analysis

In this study, we examined reasons for variability in serial intervals and identified associations of shorter serial intervals with shorter delays to case isolation and other control measures. Background Estimates of the serial interval distribution contribute to our understanding of the transmission dynamics of coronavirus disease 2019 (COVID-19). Here, we aimed to summarize the existing evidence on serial interval distributions and delays in case isolation for COVID-19. Methods We conducted a systematic review of the published literature and preprints in PubMed on 2 epidemiological parameters, namely, serial intervals and delay intervals relating to isolation of cases for COVID-19 from 1 January 2020 to 22 October 2020 following predefined eligibility criteria. We assessed the variation in these parameter estimates using correlation and regression analysis. Results Of 103 unique studies on serial intervals of COVID-19, 56 were included, providing 129 estimates. Of 451 unique studies on isolation delays, 18 were included, providing 74 estimates. Serial interval estimates from 56 included studies varied from 1.0 to 9.9 days, while case isolation delays from 18 included studies varied from 1.0 to 12.5 days, which were associated with spatial, methodological, and temporal factors. In mainland China, the pooled mean serial interval was 6.2 days (range, 5.1-7.8) before the epidemic peak and reduced to 4.9 days (range, 1.9-6.5) after the epidemic peak. Similarly, the pooled mean isolation delay related intervals were 6.0 days (range, 2.9-12.5) and 2.4 days (range, 2.0-2.7) before and after the epidemic peak, respectively. There was a positive association between serial interval and case isolation delay. Conclusions Temporal factors, such as different control measures and case isolation in particular, led to shorter serial interval estimates over time. Correcting transmissibility estimates for these time-varying distributions could aid mitigation efforts.