Global COVID-19 pandemic demands joint interventions for the suppression of future waves

Abstract

Emerging evidence suggests a resurgence of COVID-19 in the coming years. It is thus critical to optimize emergency response planning from a broad, integrated perspective. We developed a mathematical model incorporating climate-driven variation in community transmissions and movement-modulated spatial diffusions of COVID-19 into various intervention scenarios. We find that an intensive 8-wk intervention targeting the reduction of local transmissibility and international travel is efficient and effective. Practically, we suggest a tiered implementation of this strategy where interventions are first implemented at locations in what we call the Global Intervention Hub, followed by timely interventions in secondary high-risk locations. We argue that thinking globally, categorizing locations in a hub-and-spoke intervention network, and acting locally, applying interventions at high-risk areas, is a functional strategy to avert the tremendous burden that would otherwise be placed on public health and society.

Keywords: climate; disease transmission; hierarchical intervention network; human behavior; international collaboration.

Fig. 1.

Effect of seasonal climate conditions on trajectory of infection. (A) Wintertime varies across locations, that is, the longer (red) or shorter winter (green) compared with the overall duration of the winter season (yellow). Colored circles indicate the start and end of the winter season. Gray dots show the daily temperature. Dashed line is the temperature threshold for winter. (B) The cold weather in winter, through driving people indoors for a longer time, inflates the forcing on transmission. Shaded areas mark the winter seasons. Seasonal forcing suppresses the transmission by 50% and is invariant in other seasons. (C) This climate-modulated transmission risk modulates the spread of infection. Gray line shows the trajectory of infection resulting from a seasonally invariant risk of transmission, assuming $R_0\hspace{0.17em}=\hspace{0.17em}2$. The Susceptible–Exposed–Infectious–Recovered (SEIR) model is initialized with S(0) = 0.999, E(0) = 0.001, I(0) = 0, and R(0) = 0. Trajectories of infection are simulated using values for $1/\delta \hspace{0.17em}=\hspace{0.17em}5.2$ d, $1/\gamma \hspace{0.17em}=\hspace{0.17em}7$ d and $R_0\hspace{0.17em}=\hspace{0.17em}0$. Effect of seasonal forcing on transmission is modeled through $R\left(t\right)=R_0\theta \left(t\right)$, where $\theta \left(t\right)$ and $R\left(t\right)$ is the seasonal forcing and transmissibility on day t, respectively. (D) Time and magnitude of peak infections vary across locations. Peak infection is delayed and lowered in locations with a shorter duration of winter season. Circle size shows the magnitude of peak infection.

Fig. 2.

Best fit of daily temperature and seasonal forcing on community transmission. (A−F) Best fit of daily temperature in 2019 among locations in temperate, subtropical, and tropical climates in Northern (NH) and Southern (SH) Hemispheres. Wintertime in the temperate and subtropical climates is marked. (G) Locations are ordered by latitude. Magnitude of seasonal forcing is distinguished by color.

Fig. 3.

Scenario sets of intervention. We applied five intervention strategies: (A) the prompt and joint intervention among GIH locations, followed by joint intervention in secondary locations, (B) the prompt and joint intervention implemented only among GIH locations, (C) interventions initialized simultaneously among all locations, (D) the prompt and joint intervention among the GIH locations, followed by interventions in secondary locations which initialized according to location transmission risk, and (E) the intensive intervention in the GIH locations followed by a moderate intervention in secondary locations. Color distinguishes the intervention implemented among GIH (green) and secondary (yellow) locations. Circles and horizontal bars indicate the initial time of transmission and duration of intervention in each location. Dashed lines show the tiggering time of intervention.

Fig. 4.

Impact of interventions. Interventions are implemented through three scenario sets: (A and E) sequentially between GIH and secondary locations, (B and F) only in GIH locations, (C and G) simultaneously among all locations, and (D and H) sequentially between GIH and secondary, but initialized independently in secondary, locations. For each scenario set, mild, moderate, and intensive intervention reduces community transmission and international travel by 20%, 50%, and 80%, respectively. Intervention durations of 2 wk to 12 wk are considered, in increments of 2 wk. Effectiveness is evaluated in terms of the proportion of clinical cases averted and number of locations where the goal of effectively reducing incidence to <10 cases per day could be achieved in advance as compared with projections in the absence of interventions.