Key Determinants of Implementing, Lifting, and Re-Implementing the Mitigation Interventions

Backgrounds and Purpose: Social distancing is as an essential component of a public health response to infectious-disease outbreak. Therefore, the timing of enacting, lifting, and reenacting the mitigation interventions plays a vital role in fighting the pandemic. This study aims to examine the determinants associated with six types of mitigation interventions enacted, lifted, and reenacted by State Governors during the study window. We employed the following five blocks of variables in modeling social distancing orders: demographic characteristics, economic well-being information, public health infrastructure, information related to politics, and international connectivity. In addition, to measure state governmental awareness of the Covid-19, we included in the study time-varying covariates of COVID-19 counts five-days prior to the time when a mitigation policy was made.

Methods: A proportional hazards model was employed in the data analysis.

Results: States with high prevalence rates of COVID-19 (measured by the 5-day moving averages of cumulative confirmed cases per 10,000 population, the cumulative deaths per 1,000,000 population, the cumulative confirmed cases, the cumulative deaths, the new confirmed cases, and the new deaths at the time five days prior to making a decision) reacted more slowly to the outbreak. Each of these variables was statistically significant in most models analyzing the hazard rate issuing a policy. For instance, every one-case increase in the cumulative cases per 10,000 population reduces the hazard of enacting the order (i.e., in a fast speed) by 84.5% (p<.001). However, these states lifted the policy more carefully: every one-case increase in the same variable reduces the hazard of lifting the order by 0.75% (p<.05).

States with high proportions of vulnerable populations (i.e., high percentages of African Americans and Hispanics, and high unemployment rates) in general enacted the mitigation interventions earlier but lifted later. Of 52 multivariate models analyzing the timing of the Stay-at-Home order, the restaurant/bar limit, and school closure, 24 show negative impacts on the minority populations, and 6 show negative impacts on states with high unemployment rates. For instance, every one-percentage-point increase in the minority populations increases the hazard of lifting the restaurant/bar limit (i.e., in a delayed fashion when lifting) by 5.4% (p<.05). Every one-percentage-point increase in the unemployment rate reduces the hazard of reenacting the school closure (i.e., in a fast speed when reenacting) by 20.8% (p<.05).

Public health budget per capita in 2019 is a significant predictor of timing enacting the face-covering requirement. One model shows that every one-dollar increase in the budget reduces the hazard of enacting mask-wearing requirement (i.e., in a delayed fashion) by 1.7% (p<.01).

States with high proportions voting for President Biden adopted interventions earlier but lifted later. For instance, every one-percentage-point increase in the percentage voting for Biden increases the hazard of enacting the non-essential business closure by 8.4% (p<.015).

Conclusions and Implications: States with a high COVID-19 prevalence rate missed the best timing to enact the mitigation interventions. States with high proportions of minority populations and high unemployment rates started mitigations earlier and lifted later, and hence, experienced the mitigation interventions the longest time.