How we collected and analyzed the data
EDF collaborated with scientists at George Washington University to conduct a health impact assessment to understand the health impacts of air pollution in the Bay Area at a block-by-block scale. The study results and methods are described in full in a scientific paper published in Environmental Health Perspectives.
We combined fine-scale air quality data from sensors and satellites with information on the health impacts of pollutants (referred to as a concentration response function), baseline disease rates, and population data. We then developed maps that display the differences in the impacts of pollution on the health of communities across the Bay Area.
![Data analysis methodology Data analysis methodology](/sites/default/files/content/methodsVisual-670x300.jpg)
Data
Air pollution data
We used data from sensors and satellite-based models to estimate differences in air pollution concentrations across the Bay Area.
For the full Bay Area, we used estimates of annual average nitrogen dioxide (NO2) concentrations for the year 2011 at a 100 m ✕ 100 m resolution from a global model that incorporates satellite measurements, numerous land use predictor variables and annual measurement data from 5,220 air monitors in 58 countries. For black carbon and fine particulate matter (PM2.5), we used models at a 1km ✕ 1km resolution.
In areas of Oakland, California, we also utilized a large, spatially-precise dataset of mobile air pollution measurements collected as part of a previous research effort that sought to understand street-by-street variations in air pollution. In partnership with Google Earth Outreach, the University of Texas at Austin, Aclima, and the West Oakland Environmental Indicators Project, we equipped two Google Street View cars with air pollution sensors and repeatedly drove every road in West, Downtown and East Oakland, between May 2015 and December 2017, producing over three million data points. These measurements were used to quantify long term spatial differences in concentrations of NO2 and black carbon across 30m road segments. For assessing health impacts, we further aggregated the 30m segment averages to a 100m x 100m grid resolution, resulting in annual average NO2 and Black Carbon concentrations.
Concentration response functions
We evaluated health impacts for pollutants (e.g. childhood asthma attributable to NO2) where the US Environmental Protection Agency or the United Kingdom’s Committee on the Medical Effects of Air Pollutants indicate a causal or likely causal relationship between the health outcome and exposure to the pollutant. We reviewed the literature and used concentration response functions from multiple robust studies on the pollutants and health outcomes.
Baseline disease rates
The Bay Area mortality estimates employed county baseline disease rates. For pediatric asthma incidence, we applied a California state-wide baseline rate as more refined information on asthma rates in children was not available. For Alameda County, we also conducted an analysis using baseline mortality rates at the census block group scale provided by the Alameda County Public Health Department.
Population data
We used estimates of population counts from the LandScan USA dataset at 100 m resolution for the year 2017. As LandScan USA does not include age breakdowns, we calculated the fraction of the total population in different age groups using age-specific counts from the Gridded Population of the World version 4 for the year 2010, available at 1km resolution from the Socioeconomic Data and Applications Center. We estimated the proportion of people that are people of color (defined as Black, Asian, Hispanic, Pacific Islander and American Indian) for each census block group using 2010 census data.
Data analysis
To conduct the analysis, we used the concentration response functions and fine scale air pollution data to estimate the fraction of deaths and new childhood asthma cases in the area that could be attributed to the pollutant exposure. This is known as the attributable fraction.
These fractions, for both asthma and mortality, were then multiplied with the baseline rates of new childhood asthma development and mortality, respectively, in the area to estimate the number of new childhood asthma cases and deaths per year per 100,000 resulting from exposure to each pollutant. In order to evaluate the total number of people impacted, this rate was then combined with population data.