Modeling the Impact of Non-Pharmaceutical Intervention Strategies in the US During the COVID-19 Pandemic

• Activity Number: 88 - Data Visualization in Practice
• Type: Contributed
• Date/Time: Monday, August 9, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Section on Statistical Graphics
• Author(s): Tracy Jiayi Wang* and Annika J Avery and Melody Xinyu Ma and Alvin Qingkai Pan and Ben Zongyuan Yuan and Rebecca Nugent and Dr. Seema Lakdawala
• Companies: Carnegie Mellon University and University of Pittsburgh and Carnegie Mellon University and Carnegie Mellon University and Carnegie Mellon University and Carnegie Mellon University and University of Pittsburgh
• Keywords: COVID-19; Time Series; Non-pharmaceutical interventions; NPI; Lagged effect; Restrictions

Abstract

A wide diversity of non-pharmaceutical interventions (NPI) such as mask mandates and restaurant restrictions have been employed by U.S. states to combat the COVID-19 pandemic. In this work, we characterize the effectiveness of NPIs on COVID-19 transmission rates under the heterogeneity of U.S. state interventions. From March 2020 to March 2021, we scraped state and county government websites for daily information pertaining to the restricting, easing, or lifting of NPIs and then scored the NPIs by their stringency. Analyzing the autocorrelation between interventions and daily case counts, we find about a four week time lag between the NPI change and the effect on counts. To further explore the impact of different types of NPIs across states, we condition on intervention type regardless of initial date and analyze state trends from 14 days prior to the NPI change until 100 days after. We observe that states experienced similar trends after the onset of specific NPIs such as restaurant and bar restrictions. This talk will give an overview of our pipeline, data analyses, interactive dashboard maintained at phightcovid.org, and suggestions to help prepare for future pandemics.