How Can We Solve Our Air Quality Problem in the Face of Climate Change?
JAMA Network Open
Climate change is on course to disrupt nearly every aspect of the way we live, as the Intergovernmental Panel on Climate Change makes clear to us every few years. 1 Of the many impacts we can expect on human health, those mediated by changing air pollution concentrations are especially well documented. 2,3 Air pollution and climate change are intimately connected, with impacts in both directions. Climate-active pollutants, such as CO 2 , black carbon, and methane, either directly harm health or
... come from sources that emit pollutants that harm health. In addition, weather and climate have profound impacts on the levels and spatial patterns of air pollution to which we are exposed. Research over the past 2 decades has confirmed that, holding all else constant, climate change tends to worsen air pollution problems, a concept that Jacob and Winner 4 described as the "climate penalty." This effect is most pronounced for ground-level ozone, which forms on warm sunny days downwind of regions with high emissions of nitrogen oxides and volatile organic compounds (eg, metropolitan areas). Particulate matter smaller than 2.5 μm in aerodynamic diameter (PM 2.5 ) is also affected by climate change, but in more complex ways. The climate penalty means that, for any given level of air pollution emissions, pollution exposures and resulting health impacts will be made worse in a warming climate. 5 What has been less clear is how much leverage we will have during this century to mitigate the climate penalty in the continental US via more stringent air pollution control policies. This is the question that Fann and colleagues address. 6 Their approach follows a health impact assessment framework, which computes projected changes in air pollution-attributed deaths as a function of changes in air pollution levels over the current century, driven by changes in both climate and air pollution emissions. They start by running a complex air pollution simulation model over the US in 5 decadal periods, centered around 2005, 2030, 2050, 2075, and 2095. Outputs from that model are then fed into health impact functions, drawn from epidemiological studies, that relate changing air pollution concentrations to changing death rates. The authors appropriately focus on PM 2.5 and ozone, the 2 regulated air pollutants that continue to reach unhealthy levels in many parts of the world, 7 including the US. High ozone and PM 2.5 levels have consistently been associated with a wide range of adverse health outcomes, including premature death. Holding air pollution emissions constant at 2011 levels, the authors 6 estimate that a rapid climate change scenario could lead to approximately 25 000 additional pollution-attributed deaths annually by 2095, compared with the 2005 baseline year. This case isolates the health impacts of the climate penalty on air pollution as it plays out over the century. By contrast, when air pollution emissions are reduced following already-promulgated regulatory actions through 2040, projected mortality impacts are reduced by approximately 40% in 2095, to just under 16 000 additional deaths compared with 2005. These results demonstrate the substantial health benefits of additional investments in air pollution emission reduction policies. More troubling, however, is the estimate that the climate penalty will still result in additional air pollution-attributed deaths in 2095 vs 2005 even if air pollution emissions are substantially reduced. To be sure, substantial uncertainties underlie complex modeling studies of this kind. Different climate models (Fann et al 6 include 2) offer markedly different views of the timing, magnitude, and spatial pattern of future changes in temperature, PM 2.5 and ozone. Projections of future population, baseline age-specific mortality rates, and air pollution-health impact functions through 2095 are fraught with unknowns. Although most of these inputs are amenable to sensitivity analyses, it can be difficult in practice to play out the full range of scenarios because of the computational expense of +