Which factors had the greatest impact on the United States COVID-19 outbreak? An ecological assessment of mitigation behavior and policy contributions to reducing SARS-CoV-2 transmission in the US from September 2020 through November 2021

Introduction: United States’ jurisdictions implemented varied policies to slow SARS-CoV-2 transmission. Understanding patterns of these policies alongside individual’s behaviors can inform effective outbreak response. Methods: We estimated the time-varying reproduction number (R t ), a weekly measure of real-time transmission using US COVID-19 reported cases from September 2020-November 2021. Using two multi- level regression models, we then assessed the association between R t and policies, personal COVID-19 mitigation behaviors, variants, immunity, and social vulnerability indicators. First, we fit a model with state-level policy stringency according to the Oxford Stringency Index (OSI), a composite indicator reflecting the strictness of COVID-19 policies and strength of pandemic-related communication. Our second model included a subset of specific policies. Results: Implementation of stringent observed policies (defined by OSI) was associated with an R t reduction of 6.7- 11.6% (95% Credible Interval [CI]). In the Individual Policy Model, mask mandates had a null association with R t (95% CI: -1.5-0.2%), while other policies were associated with modest reductions: cancellation of public events 95% CI: 1.4-3.7%; restrictions on gathering sizes 95% CI: 0.1- 2.2%; stay-at-home orders 95% CI: 0.3-4.8%. The association between R t and other covariates was similar in both models. Among personal COVID-19 mitigation behaviors in the OSI Model, R t was estimated to decrease 18%-26% (95% CI) with a 50% reduction in national airline travel, 2.4-3.3% (95% CI) with a 10% reduction in local movement to recreation and retail locations, and 12-15% (95% CI) with self-reported mask use of 50%. Increased COVID-19 seroprevalence and vaccination were both associated with reduced R t , 28-32% (95% CI) and 20-23% (95% CI) reductions, if half of people had been previously infected or fully vaccinated, respectively in the OSI Model. Conclusion: SARS-CoV-2 transmission was reduced by layered measures. These results underscore the need for policy, behavior change, and risk communication integration to reduce virus transmission during epidemics. Key messages What is already known on this topic: When the United States responded to the COVID-19 outbreak, jurisdictions took varied approaches to balance economic well-being with health and safety. Yet, the impact of the pandemic was large - millions of people were infected and over a million people died – and the relative roles of policies, policy-independent behavior change, and other factors remains unclear. What this study adds: A retrospective analysis of policy implementation and SARS-CoV-2 transmission dynamics over a year and a half indicated that social behavioral change was critical for limiting transmission prior to increases in immunity due to infection and vaccination. How this study might affect research, practice or policy: While policies contributed to slowing the spread of COVID-19, the largest impact on transmission early in the pandemic was due to individual behavior change, highlighting the importance of identifying and communicating effective control practices regardless of specific policies.