Phase-wise evaluation and optimization of non-pharmaceutical interventions to contain the COVID-19 pandemic in the U.S

Abstract

Given that the effectiveness of COVID-19 vaccines and other therapies is greatly limited by the continuously emerging variants, non-pharmaceutical interventions have been adopted as primary control strategies in the global fight against the COVID-19 pandemic. However, implementing strict interventions over extended periods of time is inevitably hurting the economy. Many countries are faced with the dilemma of how to take appropriate policy actions for socio-economic recovery while curbing the further spread of COVID-19. With an aim to solve this multi-objective decision-making problem, we investigate the underlying temporal dynamics and associations between policies, mobility patterns, and virus transmission through vector autoregressive models and the Toda-Yamamoto Granger causality test. Our findings reveal the presence of temporal lagged effects and Granger causality relationships among various transmission and human mobility variables. We further assess the effectiveness of existing COVID-19 control measures and explore potential optimal strategies that strike a balance between public health and socio-economic recovery for individual states in the U.S. by employing the Pareto optimality and genetic algorithms. The results highlight the joint power of the state of emergency declaration, wearing face masks, and the closure of bars, and emphasize the necessity of pursuing tailor-made strategies for different states and phases of epidemiological transmission. Our framework enables policymakers to create more refined designs of COVID-19 strategies and can be extended to other countries regarding best practices in pandemic response.

Keywords: COVID-19 pandemic; human mobility; multi-objective optimization; multivariate time series analysis; non-pharmaceutical interventions; public health policymaking.

Figure 1

Implementation of anti-contagion policies in response to the COVID-19 over time and space in the United States. (A) Percentage of states with the COVID-19 policies being enacted over time. Two vertical dashed lines represent the start dates of COVID-19 policy deployment and vaccine distribution, respectively. (B) Maps of number of NPIs implemented by the states on three representative dates. April 19, 2020 represents the period with the most stringent lockdowns. December 13, 2020 is the day before COVID-19 vaccines were administered in the U.S.. August 18, 2021 is the last day of the observation period.

Figure 2

Temporal changes in daily new COVID-19 cases (NC), travel distance difference (TD), visitation difference (VD), instantaneous reproduction number (R_t) and policy implementation in the ten states with highest number of confirmed cases from February 24, 2020 to August 18, 2021. 7-day moving average is utilized to smooth volatile case reporting data and human mobility metrics. The start dates of policy implementation and vaccine distribution are indicated by dashed vertical lines. A horizontal line is drawn at R_t=1. If R_t is greater than 1, the epidemic is expanding at time t, whereas R_t < 1 signals that the epidemic is shrinking.

Figure 3

Pareto optimal trade-offs between human mobility (v_t) and virus transmission ($R_t^{\prime }$) in California during two phases. For existing solutions in phase 1 (A) and phase 2 (B), purple-colored spots represent optimal solutions that are connected by dashed lines to visually estimate the Pareto frontier. Candidate points with a value of $R_t^{\prime }$ larger than 2 are filtered. Pareto optimal points with a $R_t^{\prime }$ between 0.7 and 1 are enclosed by green boxes for phase 1 (A) and phase 2 (B), respectively. Corresponding optimal policy strategies after duplicate elimination are displayed in lower sub-figures with average $R_t^{\prime }$ and v_t listed in the tables beside them. For potential optimal policy strategies generated for California in phase 1 (C) and phase 2 (D), solutions with a ${\widehat{R}}_t$ less than 1.206 and 0.694 are selected, respectively. Corresponding parameters estimated for the policy types are displayed as heatmaps with predicted ${\widehat{R}}_t$ and ${\widehat{v}}_t$ listed in the right tables. Orange rectangle highlights the solutions similar to the existing optimal strategies of S7 and S3; and, coral ones highlight more optimized solutions.

Figure 4

Optimal response strategies generated by NSGA-II for 10 states with the highest number of confirmed COVID-19 cases during different phases. Strategies for each state are listed in ascending order of the average predicted $\widehat{R_t}$. If more than five optimal solutions are distilled for a certain state and phase, only the top five strategies are retained. Orange rectangles mark solutions already included in the existing strategies for the state; and, coral rectangles highlight more optimized ones.