Table 3 Summary of statistically significant positive associations between cardiovascular mortality, by effect modifier and single-lag days for pollutants‡

Pollutant	All	Gender		Race		Education
Pollutant	All	Male	Female	White	Hispanic	High school graduate	Non-high school graduate
PM2.5	3†	–	2†, 3†	3*	1†, 3†	–	0†, 3†
Elemental carbon	2†	2*	3*	2†	–	2*	3†
Organic carbon	–	–	3†	2†	–	–	3†
Nitrates	3*	–	3*	–	1†, 3†	–	0†, 3*
Sulphates	3†	0*	–	3*	1†	–	0†, 2†
Calcium	–	–	2*	–	–	–	–
Chlorine	–	–	–	–	2†	–	–
Copper	–	–	–	–	1†, 3*	–	1†
Iron	2†	–	2†, 3*	2†	–	2*	3†
Potassium	2*	–	2†	2†	1†	–	1†, 2*
Sulphur	3*	0†	2†	0*	3*	–	0†, 2†
Silicon	–	–	–	–	–	–	3†
Titanium	2*	–	2*	2†	1†	–	1†
Zinc	3†	0†, 2†	3†	3†	–	–	3†

PM, particulate matter.
*p<0.10 and †p<0.05 for difference from null effect.
‡Numbers in the table indicate which lags (of 0 to 3 days) were statistically significant. Bold indicates that the lag also produced a significantly greater effect estimate (p<0.10) in this group versus the corresponding demographic subgroup. The regression model includes time (4 df/year), 1-day lags of temperature and humidity, day of week and pollutant.