A methodological assessment of confounding in studies of long-term risks of air pollution