Introduction

Early epidemiological studies of SARS-CoV-2, the causative agent of COVID-19, identified certain occupations that were associated with a high risk for infection [1]. Throughout the United States (US), meat packing plants, produce processing facilities, and commercial farms have served as foci for local outbreaks, especially in less densely populated rural areas [2–5]. Viral transmission in these facilities, sometimes resulting in hundreds of cases and necessitating plant closures, has been well publicized [6, 7]. As meat packing plants were declared “critical infrastructure” early in the COVID-19 pandemic, employees were expected to continue working despite the elevated risk [8]. Infected workers may subsequently infect their household members, contributing to spread in surrounding communities [9]. In addition to the adverse effects on human health, these localized outbreaks can overwhelm medical and public health institutions, especially in underserved rural communities where these industries are concentrated, and threaten the food supply due to worker absenteeism from illness [10].

Various hypotheses have been proposed for the elevated risk among meat packing, farming, and produce processing workers, referred to collectively as food processing workers. These include risk factors both inside and outside the workplace. Risk factors external to the workplace include social determinants of health such as living conditions, social behaviors, and shared transportation. These factors are strongly associated with the demographic characteristics of the food processing workforce, which is composed disproportionately of Black and Latinx individuals [2]. Factors specific to the workplace include close physical proximity between workers, prolonged shift durations, and shared spaces and equipment. Food processing facilities may also provide favorable physical environments for SARS-CoV-2 transmission due to cold temperatures, low humidity, metallic surfaces, and poor ventilation [11, 12]. Despite the reported incidence of SARS-CoV-2 among workers in food processing industries, our understanding of how to reduce viral transmission in this population is limited. The Centers for Disease Control and Prevention (CDC) and Occupational Safety and Health Administration (OSHA) have released interim guidance on reducing risk in the meat and poultry processing industry, but little data are available on the effectiveness or implementation of these guidelines [13]. In addition, it remains unclear whether the elevated risk of SARS-CoV-2 infection among food processing workers is due to socioeconomic factors or specific work conditions, or both.

Given the unique challenges and considerations for the prevention of COVID-19 transmission in the food processing industry, a targeted approach is warranted to prevent future outbreaks among workers and their families. Therefore, we conducted an observational cohort study of meat packing plant, farming, and produce processing workers in North Carolina, which has one of the largest food processing industries in the US [14]. Here we describe SARS-CoV-2 seroprevalence and identify characteristics associated with prior SARS-CoV-2 infection among workers, including occupational and socio-demographic risk factors.

Methods

Statistical analysis

The primary outcome of interest was SARS-CoV-2 seropositivity. Potential explanatory variables of interest are shown in Table 2. We classified each worker as employed in meat packing, farming, or produce processing. Two-sample Student’s t-tests and chi-square tests were used to evaluate characteristics divided by industry (Table 1). In this analysis, we only directly compared the meat packing and farming industries due to sparse data in the produce processing industry. Next, we calculated unadjusted risk ratios using a log-binomial model and p-values using chi-square tests (when risk ratios were inappropriate) to examine the association between occupational risk factors and SARS-CoV-2 seropositivity (Table 2). We also calculated adjusted risk ratios using a traditional stepwise regression approach [19]. Variables significantly associated with the outcome in the unadjusted analysis were included as covariates; we also included age due to its significance in the existing literature [20]. We calculated adjusted risk ratios with and without highly correlated variables (Model 2 and 1, respectively). Finally, we evaluated worker characteristics divided by industry, using descriptive statistics, risk ratios, and chi-square tests (S1 Table). Again, produce processing workers were excluded due to sparse data. A p-value below 0.05 was considered statistically significant. Data were analyzed using Stata version 16 (StataCorp, College Park, Maryland).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. North Carolina food processing worker characteristics by industry (fall 2020).

https://doi.org/10.1371/journal.pgph.0000619.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. North Carolina food processing worker and workplace characteristics associated with SARS-CoV-2 seropositivity (fall 2020).

https://doi.org/10.1371/journal.pgph.0000619.t002

Results

From September to December 2020, we enrolled 118 participants who provided blood samples. Participants were 61.0% female and 94.9% Latinx, with a mean age of 41.9 (± 13.2) years. Sixty (50.8%) worked on farms, 48 (40.7%) worked at meat packing facilities, and 10 (8.5%) worked in produce processing. Demographics including age at enrollment, sex, and current smoking status were similar between industries (Table 1). Latinx individuals represented a modestly higher proportion of farm workers than meat packing workers (100% vs 88%, p 0.005).

The overall prevalence of SARS-CoV-2 antibodies in our study population was 50.0%. SARS-CoV-2 seroprevalence was highest among meat packing plant workers (64.6%), followed by farm workers (45.0%) and produce processing workers (10.0%). Age, sex, ethnicity, smoking status, frequency of in-person socializing, full time employment status, hours worked per week, and duration of employment were similar between seropositive and seronegative workers (Table 2).

Temperature and symptom checks before shifts were always required for 50.4% and 8.5% of workers, respectively. Consistent temperature checks were most common among meat packing plant workers (79.2%) and least common among farm workers (30.0%). No participants reported working after failing a temperature or symptom check; however, 16 (13.6%) workers reported developing symptoms such as fever, cough, or shortness of breath while working but continuing to work. Continuing to work after developing symptoms was associated with SARS-CoV-2 seropositivity (RR: 1.63, 95% CI: 1.14–2.32) and was most common among meat packing workers (Tables 1 and 2).

Regular mask wearing and eye protection at work were reported by 77.5% and 32.5% of participants, respectively, and 75.8% of workers reported washing their hands at work several times per day. Over a third (36.4%) of workers reported working in crowded conditions close enough to touch another worker without walking. Meat packing workers more likely than other industry workers to wear masks at work, wear protective clothing, use shields between workers, or have required temperature checks before shifts.

Prior COVID-19 illness, defined as a history of typical symptoms such as fever and cough in 2020, was reported by 35 participants (29.7%). Of the 59 participants with positive SARS-CoV-2 serology, 29 (49.2%) reported a history of COVID-19, compared to 6 (10.2%) of the 59 seronegative participants. Among seropositive workers, farm workers were significantly less likely than other industry workers to report a history of COVID-19 (22.2% vs. 71.9%, p<0.001).

Indoor, crowded, cold, and loud work environments were highly correlated with each other. High noise levels (defined as being loud enough to necessitate yelling to communicate) and cold work environments (defined as ambient temperature below 16°C) were significantly associated with SARS-CoV-2 seropositivity (RR: 1.83, 95% CI: 1.25–2.69 and RR: 1.58, 95% CI: 1.12–2.24, respectively) (Table 2). Adjusted for age and working despite symptoms, high noise levels were associated with 1.72 times the risk of SARS-CoV-2 seropositivity (95% CI: 1.16–2.55) (Table 2). Due to its high correlation with high noise levels, cold work environment was not included in the first adjusted model (Model 1). Notably, the association between loud environments and SARS-CoV-2 seropositivity remained significant in the subset of workers who did not work in cold environments (RR: 1.68, 95% CI: 1.02–2.77). Similar trends were observed in workers stratified by industry (S2 Table), although the association between high noise levels and SARS-CoV-2 seropositivity only met statistical significance among farm workers (RR: 1.23, 95% CI: 1.02–1.49). Shoe covers were associated with higher SARS-CoV-2 seropositivity among meat packing workers alone, and shields between workers were associated with higher SARS-CoV-2 seropositivity among farm workers alone (S2 Table).

Because working in cold environments was almost perfectly correlated with working inside, we also examined the effect of working indoors among those who did not work in a cold environment. Among workers who did not report working in cold environments, working indoors was not associated with SARS-CoV-2 seroprevalence (p = 0.79).

Discussion

The prevalence of SARS-CoV-2 antibodies in our study population was 50.0%, over five times the rate observed in most of the United States when we conducted this study in fall 2020 [21]. Meat processing workers had the highest prevalence, with almost two-thirds showing evidence of past SARS-CoV-2 infection. Approximately half of participants with positive SARS-CoV-2 serology reported a known history of COVID-19, suggesting a high rate of asymptomatic infections, corroborating other studies [22]. Farm workers were most likely to have SARS-CoV-2 antibodies without a reported history of COVID-19 illness, possibly reflecting lower access to testing. This is consistent with mass testing efforts at certain meat packing plants with known outbreaks in spring 2020, which may have resulted in higher testing rates in this population [23]. Farm workers also frequently experience non-specific respiratory symptoms and fatigue from occupational exposures that could be difficult to distinguish from COVID-19 illness [24].

Racial and ethnic minorities have suffered a disproportionate burden of COVID-19 morbidity and mortality, and similar trends have been seen among food processing workers [25, 26]. In the early months of the pandemic, 87% of known COVID-19 cases among agricultural workers occurred in racial or ethnic minorities, particularly Latinx communities [2]. Many of these workers are recent, sometimes undocumented, immigrants to the United States, lack political and social capital to advocate for worker safety, and have limited access to health care. Food processing workers are particularly vulnerable as they may face stressful and physically dangerous work, low wages, and limited benefits [27, 28]. Our results further demonstrate the disproportionate burden of COVID-19 among underserved and economically disadvantaged Latinx communities in the U.S.

To reduce SARS-CoV-2 transmission in food processing industries, the CDC and OSHA have recommended a number of policies such as mandatory mask wearing, social distancing between workers, frequent hand hygiene, and adequate sick leave policies to ensure workers are not penalized for reporting illness [29]. However, little if any enforcement exists for these guidelines. Our findings raise concern that inadequate safety guidelines in many food processing workplaces may have contributed to the high SARS-CoV-2 seroprevalence among workers, although many workers may have been infected early in the pandemic before these guidelines were enacted. Notably, 13.6% of workers reported developing symptoms concerning for COVID-19 during a shift but continued working. This finding highlights the importance of adequate sick leave policies for workers and policies to protect workers who report illness.

High noise levels appeared to be the primary occupational risk factor for SARS-CoV-2 seropositivity among workers. Workers who always worked in loud environments were at a 1·7 times higher risk of SARS-CoV-2 seropositivity. This association remained significant when limited to workers who did not work in cold environments, further supporting the importance of high noise levels as a risk factor. In addition, among farming and produce processing workers who reported always working in loud areas, we observed a SARS-CoV-2 seroprevalence comparable to that observed in the meat packing industry. To our knowledge, this is the first study demonstrating an association between loud work environments and SARS-CoV-2 infection risk. High noise levels may require workers to reduce distancing, shout, or remove face masks to communicate, increasing respiratory droplet emission. This association may not be due directly to high noise levels, as volume may be a surrogate marker of jobs that are high risk for other reasons. A cold work environment was also associated with higher risk of SARS-CoV-2 seropositivity in unadjusted analysis. A few studies have identified increased stability of the SARS-CoV-2 virus at cold temperatures, which could enhance aerosol or fomite-mediated transmission [30]. However, high noise level was associated with a greater magnitude of increased risk than cold environment in both unadjusted and adjusted analyses and is likely a more easily modifiable risk factor in the workplace. Our results suggest that mitigation strategies, such as reducing ambient noise levels, use of speakers, optimized facility engineering, and consistent mask use, could help reduce transmission risk in both food processing facilities and other high-risk work environments [31].

Our study identified an association between SARS-CoV-2 seropositivity and shoe covers among meat packing workers. One possible explanation for this association is that a lack of proper training on donning and doffing methods of personal protective equipment (PPE) may inadvertently lead to self-contamination of clothes and skin. However, it is more likely that particular meat packing jobs requiring shoe covers imparted a higher risk of SARS-CoV-2 transmission due to additional factors we were unable to measure, such as prolonged close contact between workers or less frequently sanitized work areas. Some meat packing facilities may also have implemented additional safety measures such as shoe covers after outbreaks. This is supported by the fact that the association between shoe covers and SARS-CoV-2 seropositivity was not observed among farming and food processing workers. Similarly, we observed an association between shield use and SARS-CoV-2 seropositivity among farm workers alone, which is likely also explained by unmeasured risk factors of jobs involving shields between workers.

Our study did not observe a significant association between reported mask wearing at work and SARS-CoV-2 seropositivity, but our surveys did not capture the technique with which masks were worn or mask type. Given the substantial evidence in existing literature that masks reduce the spread of SARS-CoV-2, our findings raise concern that many food processing workers may have poor quality masks or wear them with improper technique, such as below the nose [32]. Additionally, it is likely that many participants were infected several months before the study started, as there were known outbreaks at several plants in the spring of 2020, when mask hygiene may not have been as widespread in the workplace or community.

Our study has several limitations. We used a convenience sample that may not accurately reflect the true seroprevalence of past SARS-CoV-2 infections in this population. In addition, our study could not include individuals who died of COVID-19, though this is a small proportion of affected food processing workers [2, 33]. Behavioral data was collected by self-report, which is susceptible to a variety of biases such as recall and social desirability bias. Some workers may have been unwilling to participate due to concerns about their results affecting job security, although we assured participants their surveys and test results would remain confidential and did not recruit participants directly through employers to reduce this risk. It is also possible that our ELISA missed a few early cases due to antibody waning. We estimate this represents a very small number of our participants, as all participants were tested within 10 months of the first identified COVID-19 case in North Carolina, and prior studies have demonstrated durable antibody response for at least six to nine months [34, 35]. Furthermore, current behavior may not reflect accurately behaviors performed earlier in the COVID-19 pandemic, as public health guidelines evolved rapidly. Finally, our study population was limited to rural, almost entirely Latinx workers that may not be representative of food processing and farm worker populations in other states or countries.

In summary, our study identified a remarkably high seroprevalence of SARS-CoV-2 among meat packing, produce processing, and farm workers and their household members. These findings highlight the burden of the COVID-19 pandemic in this vulnerable population and the role of food processing facilities in community spread. This is the first study to examine occupational risk factors for SARS-CoV-2 seropositivity among meat packing, produce processing, and farm workers. Occupational noise appeared strongly associated with SARS-CoV-2 seropositivity, which has notable implications for policy recommendations and facility engineering. Although COVID-19 vaccines have become widely accessible in the US, understanding occupational risk factors remains important given incomplete vaccination rates, the emergence of new variants, and the likelihood of future airborne epidemics. This study also highlights the impacts of occupational disadvantage in the setting of COVID-19, experienced predominantly by ethnic minorities in a high-risk industry. Further research is warranted to investigate the effectiveness of mitigation strategies to reduce SARS-CoV-2 transmission among food processing workers.

Supporting information

Acknowledgments

We wish to thank the staff of Piedmont Health Services, the Hispanic Liaison, and the Episcopal Farmworker Ministry, who collaborated in the study. Premkumar Lakshmanane provided valuable advice on serological assays. We also thank Maggie Oberly and Suvra Mitra for their assistance with laboratory work. Most importantly, we thank the workers and their household members for their valuable participation and time in this study.