Methods: A descriptive analysis of event episodes and bivariate proportional hazards models were employed to address the research questions.
Results: 1) Stay-at-home order: 43 states took an average of 1420.05 hours (SD=2504.34) to enact the stay-at-home order. These states on average implemented the order for 1389.24 hours (SD=11.43.92). 15 states then enacted the stay-at-home advisory, and 2 states lifted the advisory by the end of the study window. 2) Public school closure: All 50 states and DC closed public K-12 schools since March 13, 2020, and the average time taking to make this decision is 98 hours (SD=65.96). All states reopened public k-12 schools prior to the Thanksgiving Day, and on average the reopening time is 3732.35 hours (SD=445.66). 3) Restaurant/bar limits: Except Kansas, all states took an average of 271.48 hours (SD=987.74) to enact the restaurant/bar limits, and the policy was effective for 1578.51 hours on average (SD=676.37). 4) Large gathering ban of more than 10 people: Except North Dakota, 49 states and DC enacted this order, with a mean time issuing the order of 468.85 hours (SD=1236.04). On average, this order was implemented for 1686.03 hours, while New Mexico did not lift the intervention by January 5, 2021. To prevent the transmission of COVID-19, 21 states re-issued large gathering ban of more than 10 people after its rescindment. 5) Non-essential business closure: 45 states closed non-essential business since the national emergency, with a mean time issuing the measure of 1115.74 hours (SD=2232.65). These states on average implemented the order for 1279.32 hours (SD=372.27). 6) Face covering requirement: 40 states enacted mask mandate, with a mean time taking to issue the mandate of 3972.77 hours (SD=2180.32). All these states, except Mississippi, maintained the intervention by the end of study window. Mississippi rescinded the order in October 2020.
The COVID-19 counts five days prior to the decision-making time are significant predictors of the Cox regression analyzing the determinants of mitigation interventions. The bivariate models show that, for instance, every one-case increase in the confirmed cases per 10,000 population decreases the hazard of enacting the stay-at-home order by 66.62% (p<.001). These findings indicate that states with high COVID-19 counts adopted mitigation interventions in a delayed fashion.
The bivariate models also show that states with high level of COVID-19 prevalence rate lifted the mitigation interventions in a delayed fashion, that is, they made this decision more carefully.
Conclusions and Implications: There exists a wide variation on the timing and pattern of implementing the mitigation interventions among all 50 states and DC.