Inhalation of particulate matter (PM) from the ambient air is strongly associated with disease (1–6), mortality (7, 8), toxicity is known to be concentration dependent (9), and is highest among the smaller particle sizes such as PM_2.5 (10, 11). Recent epidemiological studies have presented evidence that the burden of air pollution is unequally distributed, with differences along racial, ethnic, and class lines (12–14). In this commentary, we consider asthma, a persistent health inequity in the United States, the prevalence of which among African Americans was estimated to be 11.2%, compared to 7.6% among European Americans in 2019 (15).

Widespread measurement of particulate matter has expanded rapidly as inexpensive, internet-connected sensors are becoming widely available. But how will this new capacity be distributed? Historical health inequities in the United States and worldwide are often geographic, as illustrated by the phrase “the wrong side of the [railroad] tracks,” where industrial waste, air pollution and other health threats are concentrated in a less affluent (or segregated) area of a city, town, or other geographic region. We submit that the expansion of this new surveillance capacity is unlikely to include the large populations at the highest risk of the deleterious effects of poor air quality (16), perhaps due to the costs and requirement of the equipment, including stable power, Wi-Fi, and secure placement. Equitable distribution of air quality monitoring would involve longitudinal, fine-grained measurement of airborne pollutants in communities with elevated rates of diseases associated with these pollutants. We consider this as a unique opportunity for the scientific and academic community, as well as funders, to forge partnerships with community groups, citizen scientists, and state agencies to acquire and equitably distribute air quality monitoring devices, developing relationships that may have highly beneficial emergent properties (e.g., local knowledge of sources of emissions).

Ambient PM assessment is traditionally conducted by a field PM collection process with Federal Reference Methods (FRM) or EPA-approved FRM-equivalent approaches. These methods typically use a gravimetric filtration system run at a specific air flow rate for a specified run time or an indirect assessment of exposure level via particle counters using spectroscopic beams. While reliable and accurate, FRM studies for PM assessment are costly, requiring extensive preparation time and laboratory facilities to process specimens collected. Field measurements of any type can be challenging in historically high-burden areas due to a general distrust of researchers and lack of secure placement locations. The confluence of cost, access, and other logistical challenges has limited the collection of robust longitudinal air quality data in many communities. In North Carolina, high fidelity and longitudinal data is available from EPA and North Carolina DAQ stations. As there are a limited number of these stations, it is understood that substantial variation in air quality exists between stations. Researchers who wish to estimate pollutant burden in areas distant from air quality stations use interpolation techniques between stations (17, 18), remote sensing options including satellite imaging (19), or sample air quality at distinct points for a short period of time (20). These interpolations or short-duration monitoring efforts are often reasonable estimates of the air pollution burden for some areas but may under or overestimate it in others.

Health disparity research aims to provide actionable data to policymakers, who presumably use this data to develop harm reduction strategies. These data would ideally be collected in a wide area, be of high quality and granularity and would have regular and consistent measurement intervals. Data of this type could be used to measure concentrations of known carcinogens among groups experiencing health outcomes associated with exposure, evaluate policies intended to improve air quality, as well as serving as warning systems for acute spikes in emissions quality (e.g., wildfires). Emerging utilization of health disparity research includes action by citizen scientists to identify sources of pollutants, organizing and advocating for increased controls on emissions (21), and individuals using the information on air quality to adjust travel plans (22), and take steps to limit their exposure.

In this work, we present our template for collecting air quality information using new technology and reference methods, with a focus on communities at elevated risk of air-quality related pulmonary disease outcomes.

During the surveillance period in November–December 2020, the levels of measured PM2.5 by the PA-II wireless sensor and PEM units identified higher PM2.5 burden in CT13.01 compared to CT3.02 (Table 1). The average weight of the PM_2.5 collected by the PEM unit in CT 13.01 during the study period was forty-nine percentage points larger than that collected in CT3.02. Measurements from the PA-II unit during the same period also showed higher concentrations of PM2.5 in CT13.01, as well as more days spent in the “unhealthy” air quality index compared to CT3.02.

In this work, we demonstrate one approach to operationalizing the ideal of equity in the expansion of automated, local air quality measurements by centering data collection on populations that bear a disproportionate pulmonary disease burden. We apply a pragmatic and economically lean approach using a small NIH grant, low-cost sensors, state-collected hospital discharge data, publicly available U.S. Census demographic information, collaboration with local businesses and community organizations, and a simple analytical approach. Our results are aligned with previous research that have found that racially minoritized populations experience greater concentrations of PM_2.5 and poorer AQI in their neighborhoods compared to non-minoritized populations (12–14, 18, 29, 30).

Our approach was novel, in that we identified hot spots of asthma exacerbation by 5-digit zip code from billing data, then identified disparate spaces by race in the census tracts within. The contrasting of a census tract with a high proportion of African American residents with one with the highest proportion of European American residents aligns with the concept of centering data collection on populations known to be at risk of a given outcome; in this case asthma. We encountered significant challenges, primarily revolving around placing sensors in a location that had access, power, and internet services. Duration of monitoring was also curtailed after the loss of a PEM monitor, highlighting the challenges that monitors that require regular servicing (battery, filter changes) involve. Future endeavors could consider which location types, including residential, business or community groups, allow for the longest and highest fidelity of measurement. In many cases, investigators could consider whether small payments, or other forms of compensation, may be effective in retaining access to monitoring sites. We found that building a supportive partnership with community groups was critical to the success of this work.

It is important to note that this work is not designed to show a causal association between PM_2.5 concentration and asthma exacerbation. Instead, we demonstrate a template approach to data collection that avoids the trap of representativeness (31), instead pursuing a strategy of pragmatic utilization of free or low-cost resources to focus on a population that is at high risk of poor cardiopulmonary outcomes. We propose that a sensor network of this type, carefully positioned, could generate actionable data to support causal inference in pulmonary outcome inequity research. With the help of grant funders, community leaders, academic institutions and faith-based organizations, the future of air quality surveillance can be one that includes those currently experiencing its effects.

The hospital and emergency-room discharge data used in this work is available from the Cecil B. Sheps Center for Health Service Research. The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

This study was reviewed by the Institutional Review Board of the University of North Carolina at Chapel Hill and determined not to be human-subjects research (IRB #19-2185).

MR: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. JB: Writing – review & editing, Writing – original draft, Supervision, Project administration, Methodology, Formal analysis, Conceptualization. VS: Writing – review & editing, Supervision, Project administration, Methodology, Funding acquisition, Conceptualization.