A Proof-of-Concept Approach for Quantifying Multi-Pollutant Health Impacts Using the Open-Source BenMAP-CE Software Program
On this page:
Background
Air pollution risk assessments often employ effect coefficients from epidemiologic studies to quantify the public health impact of changes in air quality. Partly due to data and methodological limitations, epidemiologic studies have traditionally characterized the health risk of exposure to individual pollutants. Such single-pollutant approaches may: not fully account for the fact that populations are exposed to mixtures of pollutants; not account for synergistic or antagonistic health effects among populations exposed to multiple pollutants; yield biased estimates of individual pollutant effects due to collinearity with other pollutants when applied in a risk assessment. Multipollutant statistical approaches take into consideration collinearity by identifying mixtures of air pollutants that are often commonly emitted from specific sources. Applying a new proof-of-concept version of the environmental Benefits Mapping and Analysis Program—Community Edition (BenMAP-CE) we aim to answer two questions: (1) can effect coefficients from epidemiologic studies employing multipollutant statistical approaches, specifically joint effect models, be employed in air pollution risk assessments? And, (2) how does the procedure for quantifying population health impacts differ between the single and multipollutant context?
Methods
Two recent studies employ joint effect models with and without first-order pollutant interactions to estimate the risk of air pollutant-attributable asthma emergency department visits in Atlanta (Winquist et al. 2014) and the state of Georgia (Xiao et al. 2016). Within these studies, associations are examined in relation to short-term exposures to the criteria pollutants: ozone, fine particulate matter, carbon monoxide, nitrogen dioxide, and sulfur dioxide, along with particulate matter components, which collectively represent predefined source groupings (i.e., oxidant gases, secondary pollutants, traffic, power plant, and criteria pollutants). We use the effect coefficients and variance/co-variance matrices from these two studies along with daily air quality predictions from a photochemical transport model in the BenMAP-CE software program to perform an illustrative case study for the city of Atlanta and the state of Georgia.
Results
We find that: (1) the interaction models yield larger estimates of pollutant-attributable asthma emergency department visits, irrespective of pollutant group; (2) warm season impacts are greater than cold season impacts, irrespective of pollutant group; (3) certain groups, including the power plant and secondary pollutant groups, yield a negative number of cases. The BenMAP-CE software runtime for each multi-pollutant model was commensurate with the runtime for single pollutant models.
Conclusions
This proof-of-concept analysis suggests that air pollutant risk assessments of multipollutant exposures are indeed feasible, but are data-intensive. Future risk assessments that consider both single and multipollutant approaches have the potential to provide a more comprehensive evaluation that can inform air quality management strategies.