Causal modeling of air pollution

The mortality associated with exposure to PM2.5 and NO2 has long been described, but few studies have used causal modeling methods that can help moving from statistical association to causal inference. Researchers at the Harvard/MIT ACE Center  used three different methods to estimate the causal effect of local pollution on mortality in 135 US cities. Their results show mostly consistent estimates of mortality, especially at levels below 25 mg/m3, ranging between 0.83% and 1.7% per 10 mg/m3. These epidemiological analyses, despite their different assumptions, demonstrated a causal association of local air pollution with mortality even below the current daily PM2.5 standard