Title: Modeling the association of county-level factors with the SARS-CoV-2 instantaneous reproduction number and predicting transmission in counties across the United States
Abstract:
Objective: Local variation in SARS-CoV-2 transmission across the United States has not been well studied. The goal of this study is to examine the association of county-level factors with variation in the SARS-CoV-2 reproduction number over time, and provide a model to predict SARS-CoV-2 transmission at county level.
Background: Reproduction number (R), defined as the average number of people that will be infected by an individual who has the infection, play a vital role in predicting the evolution of an infectious disease outbreaks. However, the R most certainly varies by location and by time. At the individual level, variation in R is likely dependent on being in environments where exposure risk is high or of intense duration, such as for high-exposure workers in healthcare or mass transit settings, or for families living in densely crowded living conditions. At the community level, variation in R may also include population density (as a proxy for increased likelihood of crowded conditions), temperature and/or humidity (given its effects on viral propagation), policies such as social distancing, and number of susceptible individuals.
Methods: We proposed a two-step model under the generalized linear mixed effects model (GLMM) framework to model the instantaneous reproduction number (Rt), and study association of county-level factors, including social distancing, measured by percent change in visits to non-essential businesses; population density; and temperatures, with variation in the SARS-CoV-2 reproduction number over time. We also proposed an autoregressive GLMM model to predict the disease transmission at county level based on anticipated social distancing levels and weather patterns for a given region. We estimate social distancing levels by holding constant the last week’s social distancing measurements and we use historical averages for temperature and humidity to formulate our weather estimates. We assessed our predictive accuracy by comparing the observed case counts in a region with what our model predicted would happen. We included 747 counties with active outbreaks, representing 80% of the total U.S. population.
Results: Change in social distancing, population density, and daily temperature were associated with the rate of SARS-CoV-2 transmission within a county, as measured by estimated Rt. Our analysis indicates that of these three factors, implementation of social distancing has been the most significant in reducing transmission. In addition, the mitigating association of increased social distancing and moderate increases in daily temperature were most dramatic in more population dense counties, which had high Rt values that are consistent with higher R estimates from around the world.
Conclusions: Social distancing, lower population density, and temperate weather change were associated with a decreased SARS-CoV-2 Rt in counties across the US. These relationships can inform selective public policy planning in communities during the Covid-19 pandemic. The proposed model provided an effective tool to understand the effects of different local factors on disease transmission and our predictions were reasonably accurate over time.