In reply to petellis:
I'm always very suspicious of these numbers, in my professional capacity I see peer reviewed journals claiming 1000 fold different results for the same statistic.
Having had a quick squint at the NEJ paper linked to from the Guardian article and I can pick a dozen holes in it.
One of these is acknowledged by the author: "Reduced air pollution was only one factor contributing to increased life expectancies, with its effects overlapping with those of other factors".
This statement alone sums up the problem. Death has to be one of the most multi factorial symptoms ever, so to make a comparison between air pollution changes and death rates so always going to be 9/10ths guess work no matter how many fancy statistical tools applied.
I didn't see any reference to the occupations of the study participants either, I consider this to be a bit of an omission.
In all this I don't in any way deny that pollution has an impact on peoples health nut, I do not see how anyone can attribute these deaths directly to air pollution without a significant investigation into individual cases.
The investigation into the health of people living along heavily polluted roads is something more interesting. Looking into the health of children removes many of the variables.