Livestock plants and COVID-19 transmission

Abstract

Policy responses to the COVID-19 outbreak must strike a balance between maintaining essential supply chains and limiting the spread of the virus. Our results indicate a strong positive relationship between livestock-processing plants and local community transmission of COVID-19, suggesting that these plants may act as transmission vectors into the surrounding population and accelerate the spread of the virus beyond what would be predicted solely by population risk characteristics. We estimate the total excess COVID-19 cases and deaths associated with proximity to livestock plants to be 236,000 to 310,000 (6 to 8% of all US cases) and 4,300 to 5,200 (3 to 4% of all US deaths), respectively, as of July 21, 2020, with the vast majority likely related to community spread outside these plants. The association is found primarily among large processing facilities and large meatpacking companies. In addition, we find evidence that plant closures attenuated county-wide cases and that plants that received permission from the US Department of Agriculture to increase their production-line speeds saw more county-wide cases. Ensuring both public health and robust essential supply chains may require an increase in meatpacking oversight and potentially a shift toward more decentralized, smaller-scale meat production.

Keywords: COVID-19; agriculture; livestock; public health; supply chains.

Fig. 1.

Mean county-level COVID-19 cases per thousand (A) and deaths per thousand (B) over time based on proximity to a livestock facility. The band “0–50 km” excludes the county itself. Counties are categorized into nonoverlapping, single categories based on the nearest facility (e.g., if a county contains a livestock facility and is within 50 km of another facility outside the county, the county is coded “In county” and not “0–50 km”). A visualization map is included in SI Appendix, Fig. S1.

Fig. 2.

Shaded counties contain at least one beef or pork facility categorized by USDA FSIS as processing more than 1 million pounds per month (categories 4 and 5) or at least one poultry facility categorized as processing more than 10 million pounds per month (category 5).

Fig. 3.

Relationship between COVID-19 cases and livestock plants owned or operated by large meatpacking companies. Coefficients are firm fixed-effect coefficients plotted from SI Appendix, Table S6. Error bars represent 95% CIs.

Fig. 4.

Graphs match COVID-19 pretrends of control group (green lines) to counties with plant shutdowns (red lines) based on percent growth in cases (weekly log difference) in the 2 wk prior to shutdown. Selected counties are in the top quartile of growth rates among the 233 counties with livestock plants that did not have a plant shutdown. For nonshutdown counties, week 0 is assigned to the mean shutdown date, April 22, 2020. A and B plot coefficients from a panel regression, where counties are interacted with the weekly event index in terms of percent growth in cases (A) and change in case rates per 1,000 (B). Estimates are relative to the baseline trend across all counties. One week prior (week −1) is omitted as the reference level. Models control for stay-at-home orders at the state level and include a fixed effect for each county. Error bars reflect a 95% CI. C and D are daily line charts of the mean values of each group in terms of percent case growth and change in case rate, respectively. Gray shaded bars reflect the estimated period when the effect of closing a plant would have been reflected in cases (1 to 3 wk after), given that incubation periods may last up to 14 d) (62).