COVID-19 infection data encode a dynamic reproduction number in response to policy decisions with secondary wave implications

Abstract

The SARS-CoV-2 virus is responsible for the novel coronavirus disease 2019 (COVID-19), which has spread to populations throughout the continental United States. Most state and local governments have adopted some level of "social distancing" policy, but infections have continued to spread despite these efforts. Absent a vaccine, authorities have few other tools by which to mitigate further spread of the virus. This begs the question of how effective social policy really is at reducing new infections that, left alone, could potentially overwhelm the existing hospitalization capacity of many states. We developed a mathematical model that captures correlations between some state-level "social distancing" policies and infection kinetics for all U.S. states, and use it to illustrate the link between social policy decisions, disease dynamics, and an effective reproduction number that changes over time, for case studies of Massachusetts, New Jersey, and Washington states. In general, our findings indicate that the potential for second waves of infection, which result after reopening states without an increase to immunity, can be mitigated by a return of social distancing policies as soon as possible after the waves are detected.

Figure 1

The ERDC SEIR model forecasts for Massachusetts, New Jersey, and Washington. The black dots represent the number of active cases as represented by data, while the red line is the median number of active cases predicted by the model.

Figure 2

The ERDC SEIR second wave hypothetical scenarios for Massachusetts, New Jersey, and Washington, assuming that SIPOs are lifted on 15 July 2020. The black curves represent the median number of reported infections predicted by the model if no population immobilization policies (e.g., shelter-in-place orders) are put in place to mitigate the second wave. The red curves represent the median number of reported infections predicted by the model if shelter-in-place policies are enacted 2 weeks beyond 15 July.

Figure 3

The dynamic reproductive number for SARS-CoV-2 over the two wave scenarios for Massachusetts, New Jersey, and Washington as presented in Fig. 2. The black curves represent the reproductive numbers over time for the open scenario, while the red curves represent the reproductive numbers over time for the shelter-in-place scenarios.