Racial Disparities in Air Pollution Burden and COVID-19 Deaths in Louisiana,USA,in the Context of Long-Term Changes in Fine Particulate Pollution

Background

Louisiana has been among the U.S. states hardest hit by the COVID-19 pandemic, with 88,590 cases and 3399 related deaths reported through July 17, 2020. ¹ However, these impacts have not been evenly distributed ² ; Black Americans account for 33% of Louisiana's population, but an alarming 51% of COVID-19 deaths in the state. ³ Similarly, disproportionate numbers of COVID-19 hospitalizations/deaths among Black Americans have been reported in multiple cities and states, and across the entire United States. ⁴ These disparities likely result from a complex mix of biological, social, and environmental factors, ⁵ and air pollution is predicted to be one such factor. ⁶ Identifying potential links among race, pollution, socioeconomics, health, and COVID-19 deaths in Louisiana may help elucidate, and ultimately address, the intractable public health disparities in this state.

A growing body of research provides compelling evidence that exposure to common air pollutants increases susceptibility to respiratory diseases, including from viral infections. ⁷ For example, exposure to particulate matter (PM) has been linked to higher rates of chronic cough, bronchitis, and chest illness among U.S. schoolchildren, ⁸ as well as increased hospital admissions for pneumonia in both children and adults. ⁹ Several studies suggest that even short-term (<7 days) increases in PM can result in higher rates of respiratory infections among children and adults. ¹⁰ Both short-term and long-term measurements of fine PM_2.5 exposure were linked to higher death rates from SARS, a coronavirus, during the 2003 outbreak in China. ¹¹ Experimental studies are beginning to elucidate the mechanisms underlying these associations, with evidence that PM suppresses the early immune response, for example by reducing macrophage activity. ¹²

Given the well-established link between air pollution and respiratory disease, there is growing concern that air pollution may increase susceptibility to COVID-19. This disease, caused by a novel coronavirus (SARS-CoV-2) that attacks the lungs, is characterized by respiratory distress and pneumonia. ¹³ Indeed, air pollution exposure is associated with many of the co-morbidities that increase risk of severe illness or death from COVID-19, including diabetes, hypertension, lung cancer, asthma, and chronic obstructive pulmonary disease. ¹⁴ A recent, nationwide analysis from researchers at Harvard University found that a small increase in PM_2.5 exposure (measured over the short term or long term) was associated with a large increase in per capita COVID-19 death rates. ¹⁵ The analysis, currently under peer review, included more than 3000 counties and accounted for 20 confounding factors (e.g., diabetes, obesity, and days since first reported case [18]). At least eight other studies (representing the U.S., China, and several European countries) have reported geographic or temporal patterns that are consistent with an association between air pollution and COVID-19 death rates, including a robust analysis of 355 municipalities in The Netherlands that reported a nearly 100% increase in COVID-19 cases for every 20% increase in PM_2.5 concentrations. ¹⁶ Although most of these studies have yet to undergo peer review, they collectively provide compelling evidence for a link between air pollution exposure and increased COVID-19 susceptibility.

Concerns about a possible link between air pollution and COVID-19 have brought national attention to Louisiana's “Cancer Alley,” a heavily industrialized, heavily polluted area with high percentages of Black Americans and high poverty rates. ¹⁷ Also referred to as the “Industrial Corridor” or “Death Alley,” this ∼130 river-mile, winding stretch of land along the Mississippi River encompasses more than 200 industrial sources of air pollution from Baton Rouge to New Orleans, including petrochemical facilities, metal manufacturers, and fertilizer plants. ¹⁸ In response to the evidence linking air pollution and COVID-19, the Louisiana Chemical Association claimed that “growing industry is actually causing emissions to go down, not up” and that vehicle emissions were a major PM source in Louisiana. ¹⁹ This narrative contradicts the firsthand experiences of Cancer Alley residents, many of whom have a higher risk of cancer from air pollution than ≥95% of Americans. ²⁰ Environmental justice efforts have been ongoing in Louisiana for decades, as petrochemical development has continually targeted African American communities, ²¹ many of which were founded by enslaved people after emancipation. ²² For these communities, “victories” are hard-fought and rarely (if ever) achieve true justice, as illustrated by grassroots campaigns for paid relocation—a goal that would dissolve a community to protect the health of its citizens. ²³ Louisiana's environmental justice efforts have intensified in response to a new, ongoing wave of industrialization that began in ∼2014, largely driven by U.S. natural gas production. ²⁴ Although disjunct from Cancer Alley, Calcasieu Parish (southwest Louisiana) is, in a sense, the epicenter of the Louisiana's environmental justice efforts, with the highest industrial emissions in the state by a wide margin. ²⁵

In both Cancer Alley and southwest Louisiana, there are major gaps in the air monitoring network maintained by the Louisiana Department of Environmental Quality (LDEQ) to determine compliance with federal air quality standards. For example, LDEQ has no PM_2.5 monitors in five of the 11 Cancer Alley parishes (Ascension, Plaquemines, St. Charles, St. James, and St. John the Baptist), four of which also lack a monitor for nitrogen oxides (NO_x), another type of pollutant of major concern for respiratory health. ²⁶ Although these “gold-standard” monitors are needed to ensure that air quality meets federal standards, pollution disparities can be identified by using data from emissions inventories ²⁷ or satellite measurements. ²⁸

From an environmental justice perspective, the existence of a pollution disparity is sufficient to warrant corrective action, regardless of the cause(s) of that disparity. Specifically, according to EPA, environmental justice requires that “no group of people, including a racial, ethnic or a socioeconomic group, should bear a disproportionate share of the negative environmental consequences from industrial, municipal and commercial operations or the execution of federal, state, local and tribal programs and policies.” ²⁹ Thus, basic, exploratory research (including the study reported here) is needed to identify environmental disparities, independent of understanding their underlying driver(s). However, no study (to our knowledge) has evaluated the existence of pollution disparities in Cancer Alley since 2013, when toxic facilities were reported to be disproportionately located in Black and low-income communities (7). To our knowledge, the only other study to examine pollution disparities in Cancer Alley used data from the 2005 National Air Toxics Assessment and found that Black American and low-income communities are overburdened with air pollution. ³⁰

Here, we explore the relationships among pollution, race, socioeconomic status, and per capita COVID-19 death rates in Louisiana (through July 17, 2020) and examine long-term, statewide trends in PM_2.5 pollution. We present two parallel disparity analyses, one at the census tract level (excluding COVID-19 data) and one at the parish level (including COVID-19 data, which are unavailable at a finer geographic scale). For both analyses, we use two independent measures of pollution burden: satellite-derived PM_2.5 estimates ³¹ and EPA estimates of health risks (i.e., respiratory hazard [RH] and immunological hazards [IHs]) from the combined effects of all modeled air pollutants. ³² We chose these hazard indices given their relevance to COVID-19 susceptibility. ³³ We predicted that pollution burden would be positively correlated with the percentage of Black residents in the population and with measures of economic hardship. Further, we predicted that parish-level per capita COVID-19 death rates would be positively correlated with pollution burden, percentages of Black residents, and economic hardship. Our goals were to identify any existing pollution/COVID-19 disparities and determine whether Louisiana's PM_2.5 pollution levels (from industrial and nonindustrial sources) have changed over the past several decades.

Methods

Results

Relationships among air pollution burden, race, and poverty in Louisiana at the census tract level

Louisiana census tracts with larger percentages of Black residents had higher pollution burdens (PM_2.5, RH, spRH, and IH; Fig. 1 and Table 1). Higher unemployment rates also were associated with increased pollution burden (PM_2.5, RH, and IH), whereas poverty was negatively correlated with some measures of pollution burden (RH and IH; Table 1). Conversely, most or all measures of pollution burden were lower in census tracts with larger percentages of whites or larger percentages of senior citizens, respectively (Table 1).

OPEN IN VIEWER

FIG. 1.

RH, IH, and percentage of Black American residents by census tract. Hazards are from the most recent (2014) National Air Toxics Assessment and reflect stationary point sources of pollution only (i.e., excluding vehicles and other mobile or nonpoint sources). Hazards are broken down by quantile. IH, immunological hazard; RH, respiratory hazard.

OPEN IN VIEWER

Table 1.

Spearman's Correlations Among Pollution, Race, and Poverty in Louisiana at the Census Tract Level

Demographic variable	Pollution estimate
	Long-term pm2.5 a		Respiratory hazard b				Immunological hazard b
			All sources		Point source		All sources		Point source
	rho	p	rho	p	rho	p	rho	p	rho	p
% Black	0.23	<0.0001	0.11	0.0004	0.07	0.014	0.16	<0.0001	0.05	0.098
% White	−0.27	<0.0001	−0.07	0.016	−0.09	0.002	−0.19	<0.0001	NA	0.13
% 65 years+	−0.14	<0.0001	−0.10	0.0002	−0.06	0.042	−0.15	<0.0001	−0.08	0.006
% Unemployed	0.12	<0.0001	0.07	0.019	NA	0.61	0.07	0.013	NA	0.65
% Poverty	NA	0.12	NA	0.60	−0.15	<0.0001	−0.05	0.074	−0.08	0.008
% Seniors in poverty	0.05	0.079	NA	0.22	−0.13	<0.0001	−0.08	0.007	−0.08	0.005

Statistically significant (p < 0.05), positive associations are emphasized with red text; rho = Spearman's coefficient; p = significance value. (Higher absolute rho values indicate a stronger linear relationship, whereas lower p-values indicate a more statistically significant relationship.)

a

Mean of annual ground-level concentrations from 2000 to 2016; n = 1079 census tracts.

b

From the 2014 National Air Toxics Assessment; n = 1126 census tracts.

NA, not applicable because the relationship is nonsignificant, and there is no tendency toward significance (i.e., p > 0.10).

COVID-19 death rates relative to pollution estimates, race, and poverty in Louisiana at the Parish level

With two exceptions (Tensas and Cameron parishes), every parish in Louisiana had reported at least one COVID-19 related death by July 17, 2020. Six of the 10 parishes with the highest COVID-19 death rates in Louisiana through July 17, 2020 were in Cancer Alley: St. John the Baptist, St. James, Orleans, Iberville, West Baton Rouge, and Jefferson (Table 2). Among these parishes, COVID-19 death rates were 2.1- to 3.7-fold above the state median; by contrast, only 4 of the 53 parishes outside Cancer Alley were in this top 10 group, 2 of which were adjacent to Cancer Alley Parishes (Table 2). Across all parishes, increased per capita COVID-19 death rates were associated with higher spRH, IH, and spIH (Table 3 and Fig. 2A, B). In addition, increased COVID-19 death rates tended (p = 0.088) to be associated with higher long-term (2000–2016) PM_2.5 concentrations (Table 3). Higher death rates also were associated with higher percentages of Black residents (Fig. 2C), greater population density, and more days since the first reported COVID-19 case among parishes (Table 3). No other factors were significantly associated with per capita COVID-19 deaths among parishes, including the prevalence of diabetes and poverty (Table 3 and Appendix A2).

OPEN IN VIEWER

FIG. 2.

Geographic distribution of per capita COVID-19 death rates (reported through July 17, 2020), relative to RH (A), IH (B), and percentage of Black American residents (C). Hazards are from the most recent (2014) National Air Toxics Assessment and reflect stationary point sources of pollution only (i.e., excluding vehicles and other mobile or nonpoint sources). Hazards are broken down by quantile.

OPEN IN VIEWER

Table 2.

Parishes with the 10 Highest Per Capita COVID-19 Death Rates in Louisiana, as of July 17, 2020

Rank	Parish	COVID-19 deaths per 10,000 population	X-fold above state median	Cancer Alley parish?
—	State median	5.5	—
1	Bienville	21.2	3.8	No
2	St. John the Baptist	20.5	3.7	Yes
3	East Feliciana	17.8	3.2	Adjacent
4	St. James	14.2	2.6	Yes
5	Orleans	13.9	2.5	Yes
6	Union	13.6	2.5	No
7	Iberville	13.2	2.4	Yes
8	Pointe Coupee	12.9	2.3	Adjacent
9	West Baton Rouge	12.1	2.2	Yes
10	Jefferson	11.5	2.1	Yes

Out of 64 parishes in Louisiana (n = 11 Cancer Alley parishes).

OPEN IN VIEWER

Table 3.

Pearson Correlations with Parish-Level per Capita COVID-19 Death Rates, as of July 17, 2020

Factor	r	P
Pollution-related
Long-term mean PM2.5	0.21	0.088
RH—all pollutant sources	0.22	0.082
RH—stationary point sources	0.46	0.0001
IH—all pollutant sources	0.43	0.0004
IH—stationary point sources	0.35	0.005
Demographic
Percent Black	0.32	0.011
Percent White	−0.34	0.006
Percent 65 years+	NA	0.79
Socioeconomic
People in poverty (%)	NA	0.82
Seniors in poverty (%)	NA	0.97
Unemployment rate (%)	NA	0.97
Health
Diabetes (% of adults aged 20+)	NA	0.69
Obesity (% of adults aged 20+)	NA	0.48
Smoking (%) a	−0.28	0.026
Other
Population density	0.37	0.003
Days since first case	0.47	<0.0001

n = 64 parishes; statistically significant (p < 0.05) positive associations are emphasized with red text; r = Pearson's coefficient; p = significance value. (Higher r values indicate a stronger linear relationship, whereas lower p values indicate a more statistically significant relationship.)

a

Analyzed using Spearman's correlation, as data were non-normally distributed and transformation did not improve normality.

IH, immunological hazard; NA, not applicable because the relationship is nonsignificant, and there is no tendency toward significance (i.e., p > 0.10); PM, particulate matter; RH, respiratory hazard.

Long-term changes in emissions and ground-level concentrations of PM_2.5 in Louisiana

Annual PM_2.5 emissions (all sources combined) in Louisiana decreased 17% overall from 1990 to 2017, based on National Inventory Trends Data (Fig. 3A). Total industrial PM_2.5 emissions were similar between 1990 and 2017, but with large variations in the interim (Fig. 3B). Some of this fluctuation may be due to reported changes in EPA's estimation methodology. ⁴² Vehicle PM_2.5 emissions decreased progressively from 1990 to 2017, dropping 75% overall (Fig. 3B). Fuel combustion from electric utilities decreased overall by 67%, whereas fuel combustion from other sources (i.e., residential, commercial, and institutional) gradually declined by 68% across this period (Fig. 3B). Industrial sources accounted for 20% of Louisiana's total PM_2.5 emissions in 1990, but they accounted for 25% in 2017 (Fig. 3A). From 2014 to 2017, overall emissions declined by 20%, vehicle emissions declined by 29%, and emissions from nonindustrial fuel combustion declined by 18%, whereas industrial PM_2.5 emissions increased by 33% (Fig. 3B).

OPEN IN VIEWER

FIG. 3.

Emissions of PM_2.5 over time by source category with (A) and without (B) total emissions from all sources included. Miscellaneous sources are omitted from these figures (but included in the calculation of total emissions) to better illustrate the relative contributions of these source categories. See methods for more information on source categories. Values from the 2017 National Emissions Inventory Trends dataset. *Data unavailable for 1991–1995. PM, particulate matter.

Surface-level PM_2.5 concentrations declined statewide from 2000 to 2015, based on modeled satellite data (Fig. 4A). These concentrations subsequently increased from 2016 to 2018 across southern Louisiana (Fig. 4A). From 2000 to ∼2015, Louisiana's major cities were PM_2.5 pollution hotspots (from south to north): New Orleans, Lake Charles, Lafayette, Baton Rouge, Alexandria, Shreveport, and Monroe (Fig. 4 A, B). In 2017 and 2018, the highest levels of PM_2.5 pollution occurred over a broad swath of south Louisiana and were no longer concentrated in cities (Fig. 4A).

OPEN IN VIEWER

FIG. 4.

Levels of PM_2.5 in Louisiana over time, relative to the overall U.S. long-term average (8.4 μg/m ³ ; from Wu et al., 2020), with the upper limit of the scale set to illustrate changes over time (A) or hotspots in cities (B).

Discussion

Our analysis yielded three major findings. First, Black communities and economically disadvantaged communities in Louisiana continue to be overburdened with pollution and the associated health risks. Second, increased per-capita COVID-19 death rates among Louisiana parishes are associated with higher estimates of pollution burden and larger percentages of Black residents. In other words, Louisiana's communities of color have long been overburdened with air pollution and are now overburdened with COVID-19. These COVID-19 associations were not driven by the prevalence of diabetes, obesity, or smoking. Finally, Louisiana's patterns of PM_2.5 pollution have changed dramatically over time, with vehicle emissions declining substantially, industrial sources becoming a relatively larger fraction of PM_2.5, and the state recently losing ground on long-term air quality improvements.

Our findings are consistent with previous research demonstrating racial and economic disparities in pollution burden in Cancer Alley, using other datasets. Specifically, James et al. (2012) demonstrated that air pollution in this region disproportionately affects socially disadvantaged and racial minority communities, using Cancer Risk estimates from the EPA's 2005 National Air Toxics Assessment (NATA). Perera and Lam (2013) reported that Cancer Alley communities located close to toxin-emitting facilities were disproportionately minority and disproportionately impoverished, using EPA Toxic Release Inventory data. In this study, we relied on long-term PM_2.5 concentrations estimated from satellite data, ⁴³ as well as on RH Index and IH Index from the EPA's most recent (2014) NATA. Thus, there are now at least three independent datasets that provide evidence of racial and socioeconomic disparities in pollution burden in Louisiana. Although the correlation coefficients reported here are generally low, these values indicate the strength of the linear relationship between two variables. It is important to consider that these relationships may be more complex than a simple linear association. For example, the relationship between pollution and race may be different for heavily polluted areas compared with moderately polluted areas. Fully characterizing the nature of these relationships was beyond the scope of this exploratory analysis, but it is a priority for future research and may provide insight into the drivers of pollution and COVID-19 disparities in Louisiana. Further, these relationships should be explored by using finer-scale COVID-19 data (e.g., from census tracts), once that information becomes available from the Louisiana Department of Health.

Understanding temporal changes in pollutant concentrations and the relative contributions of different emission sources may help address pollution disparities. We found that, overall, concentrations of PM_2.5 have declined substantially in Louisiana overall the past two decades (Fig. 4A). This decline has manifested in two ways: the disappearance of PM_2.5 hotspots from cities and lower PM_2.5 levels across the entire state (Fig. 4 A, B). Contributing factors likely include the decommissioning of coal-fired power plants in the broader region ⁴⁴ and the observed 75% reduction in vehicle emissions since 1990 (Fig. 3). By contrast, industrial emissions remained relatively unchanged in 2017 compared with 1990, despite substantial variation in the interim (Fig. 3). Thus, we found no evidence that industrial sources are responsible for the steady decline in PM_2.5 concentrations observed in Louisiana from 2000 to 2015.

Lowered PM_2.5 concentrations have not been maintained in all regions of Louisiana; the southern part of the state (including Cancer Alley) has, in the last couple of years, lost ground on long-term improvements in PM_2.5 (Fig. 4). Although identifying the reasons for this trend is beyond the scope of our study, we note that industrial PM_2.5 emissions (from all sources combined) increased by a third from 2014 to 2017, the most recent year for which data are available (Fig. 3). Regardless, our findings directly contradict the Louisiana petrochemical industry's current focus on vehicle emissions as a major source of PM_2.5 pollution in Louisiana ⁴⁵ and refute the narrative that “growing industry is actually causing emissions to go down, not up.” ⁴⁶ Importantly, our study indicates that pollution disparities in Louisiana are not simply the result of vehicle emissions or urbanization, because disparities persisted when pollution data (RH) were limited to stationary point sources (e.g., excluding vehicles and residences; Tables 1 and 3). Importantly, health risks from stationary point sources (i.e., spRH and spIH) also were associated with higher per capita COVID-19 death rates (Table 3). Collectively, these findings are consistent with well-publicized concerns that toxin-emitting industrial facilities represent a serious health threat and are disproportionately impacting impoverished Black communities in Cancer Alley. ⁴⁷

Although Cancer Alley communities and their allies have been fighting for environmental justice for decades, the COVID-19 pandemic highlights the urgency of this issue. Louisiana's pollution disparities are evidenced by multiple independent data sources, including those reported here, and it is imperative that environmental decision makers and public health professionals in Louisiana acknowledge and address these inequities. Central to this goal is the need for more extensive air quality monitoring in Louisiana. The current network of monitors operated by the LDEQ is grossly inadequate. For example, LDEQ monitors PM_2.5 levels at only 20 sites across the state (an area of more than 52,000 square miles). ⁴⁸ The lack of air quality data in Louisiana is particularly conspicuous in Cancer Alley, with only one PM_2.5 monitor for determining National Ambient Air Quality Standards (NAAQS) compliance along the ∼130 river-mile stretch of the Mississippi River in between Baton Rouge and Jefferson Parish. ⁴⁹ Public access to reliable, continuous local air quality data is essential to promoting public health and environmental justice in vulnerable communities. Our study emphasizes the value of this information, as well as the need to consider long-term exposure to certain pollutants (i.e., those posing RHs or IHs) as a risk factor for respiratory disease, including COVID-19. Framing pollution in this context will help advance public health efforts for communities worldwide, including those in Cancer Alley, that are overburdened with pollution.