The COVID-19 pandemic has revealed the importance of disease surveillance data for understanding the number of COVID-19 cases and deaths within communities, and how these outcomes are tracking over time. Increases or decreases in diagnosed cases over time can signal the failure or success of disease control efforts, and guide policy-makers when making decisions about tightening or relaxing social distancing policies. While of great importance for understanding disease trends, surveillance data are typically affected by reporting lags, which can greatly reduce the utility of these data and make it seem like epidemics are declining even as the true number of cases continues to increase. In a seminar for the Harvard T.H. Chan School of Public Health, Assistant Professor Nick Menzies described some of the approaches being used to deal with reporting delays in COVID-19 surveillance data. These ‘nowcasting’ methods fill in cases missing from the raw surveillance data – disease cases that have presented for diagnosis but have not yet been diagnosed and reported into the surveillance system. Menzies profiled a nowcasting software package—‘NobBS’—developed by a Harvard Chan graduate Sarah McGough along with Harvard Chan Professors Menzies and Marc Lipsitch, and Michael Johansson of the U.S. Centers for Disease Control and Prevention. This package allows users to estimate the true trend in COVID-19 cases from surveillance data affected by reporting delays, and has been used by public health agencies in the United States and internationally to track trends in COVID-19 cases and deaths. Menzies also described extensions being made to these approaches to understand the changes in transmission and incidence associated with reported COVID-19 outcomes, and how these outcomes correlate with evidence on social distancing behaviors.
Learn more: Read the publication, Nowcasting by Bayesian Smoothing: A Flexible, Generalizable Model for Real-Time Epidemic Tracking
Related news: Valuing COVID Mortality Risks