You can't randomize people to breathe dirty air for 20 years. That forces researchers into observational studies where hazard ratios for pollution hover around 1.3 — far too small for epidemiology to establish causality. Smoking gave us hazard ratios above 10; pollution rarely comes close.