Supplement Table 1. Limit of detection and precision for each trace element analyzed by ICP-MS
Element / Isotope / Method Detection Limit* (ppb) / Precision^ (%)Na / 23 / 1.845 / 4.0
Mg / 24 / 0.528 / 2.1
Al / 27 / 0.053 / 4.0
P / 31 / 0.235 / 5.0
S / 32 / 8.305 / 3.9
K / 39 / 0.229 / 1.8
Ca / 44 / 4.998 / 4.6
Ti / 47 / 0.006 / 0.4
V / 51 / 0.020 / 4.0
Cr / 52 / 0.054 / 4.5
Mn / 55 / 0.046 / 3.4
Fe / 57 / 0.082 / 4.2
Co / 59 / 0.019 / 3.2
Ni / 60 / 0.042 / 4.3
Cu / 63 / 0.084 / 9.3
Zn / 66 / 0.074 / 7.0
As / 75 / 0.008 / 1.2
Se / 77 / 0.058 / 0.1
Rb / 85 / 0.001 / 2.3
Sr / 88 / 0.402 / 1.8
Mo / 95 / 0.153 / 9.9
Ag / 107 / 0.009 / 2.2
Cd / 111 / 0.016 / 2.2
Sn / 118 / 0.005 / 0.4
Sb / 123 / 0.010 / 1.9
Ba / 137 / 0.173 / 0.6
La / 139 / 0.000 / 1.4
Ce / 140 / 0.000 / 0.9
Pb / 208 / 0.025 / 3.1
*Three times the standard deviation of seven consecutive measurements of a spiked blank
^Based on replicate analyses
Supplement Table 2. Source factor contributions in %.
Samples used for PMFFactors / 138 samples with OC/EC data / 138 samples without OC/ EC data / 69 samples
Refinery / 0.1 / <0.1 / 1
Metal processing/ incinerator / 3 / 5 / 7
Secondary / 51 / 50 / 60
Mobile / 18 / 15 / 14
Diesel/dust / 12 / 14
Iron/Steel / 9 / 5 / 12
Unidentified / 7 / 11 / 6
Supplement Table 3. Correlation matrix for PM2.5 mass and chemical components.
Supplement Table 4. Associations between chemical components and same-day changes in health outcomes.
Outcome / Component / β estimate* / PSBP / Ba / -1.097 / 0.061
Ce / -1.287 / 0.061
Mg / -1.065 / 0.096
Ti / -1.299 / 0.042
Fe / -1.026 / 0.056
DBP / Al / -0.764 / 0.093
AIx@75 / Co / 3.754 / <.0001
PWV / Sb / 0.100 / 0.059
Ba / 0.210 / 0.004
Ni / 0.101 / 0.025
HF / Mo / 519.5 / 0.0001
LF / Rb / -130.1 / 0.026
Mg / -155.5 / 0.060
Ti / -147.6 / 0.076
Mn / -145.7 / 0.069
SBP, systolic blood pressure; DBP, diastolic blood pressure; AIx@75, augmentation index at a HR of 75 beats per minute; PWV, pulse wave velocity; HF, high frequency HRV power; LF, low frequency HRV power.
*Mixed model association of outcome changes per IQR increase in metal concentration during the same-day exposure period. Only results with p values < 0.1 are shown.
1
Supplemental Material:
Non-Invasive Human Vascular Function Studies
Many conditions can affect resting arterial compliance and vascular reactivity (i.e. microvascular endothelial-dependent vasodilatation). It was essential to standardize conditions as much as possible among the measurement days during all testing. To avoid diurnal variations, the time of day for starting the vascular study protocols remained within the same 2-hour window for each individual and was performed in the same dedicated research vascular laboratory room after calmly resting supine on a patient examination bed for 10 minutes. Room temperature was monitored and kept constant (70-72 °F). All medications, over-the-counter drugs, or herbs that potentially affect vascular reactivity were discontinued prior to enrollment (or kept constant throughout the study period without changes). The same vascular technician performed all repeat testing on individual subjects within a study.
After resting supine for 10 minutes, 3 supine blood pressures (BP) using a semi-automated Omron 780 were measured on the non-dominant arm 30 seconds apart and the results were averaged. The order of additional vascular protocol testing was: heart rate variability (HRV) analyses, arterial waveform analyses and pulse wave velocity for compliance (SphygmoCor) and then measurement of microvascular endothelial-dependent vasodilatation by the EndoPat2000 device.
Heart Rate Variability
The basis for heart rate variability (HRV) assessment in the time and frequency domain of Holter monitor-derived ECG signals is that it provides a measure of cardiac autonomic nervous system activity and balance between the sympathetic and the parasympathetic nervous system tone. HRV is a standard, validated method to assess autonomic balance, particularly in studies related to air pollution health effects1.
Continuous ECG Holter monitoring was performed using a digital 4-lead “evo Holter monitor” (by Spacelabs, formally Delmar) ( Each subject’s skin was prepared and dried and electrodes were placed in a standard 4-lead position and attached to the Holter device. Holter ECG monitoring took place after subjects rested supine for at least 10 minutes. The period of Holter ECG monitoring for all HRV analyses typically consisted of a 10-minute epoch of recording, although in some circumstances analyses of 24 hours (only for time domain analyses) can also be performed.
Data was analyzed utilizing the Spacelabs software package for HRV (Pathfinder) ( provided by the company. The time domain HRV principally reflects alterations in vagal balance1. A reduction in time domain, standard deviation of all normal RR intervals (SDNN) served as the usual primary outcome for studies, as it has been directly linked with a worse CV prognosis1. Time domain metrics can be measured and provide an integrated result for periods as short as 5 minutes and as long as up to 24 hours. Other time domain outcomes (RMSSD, PNN50) were also obtained. Frequency domain analyses (5-6 minute period time period epochs) were also determined (total, low, and high power spectral density; low and high frequency normalized power; low/high frequency ratio). These analyses represent the balance (low/high frequency ratio) between the predominantly vagal (high frequency) and predominately sympathetic (low frequency) nervous system activities1.
Reference
1. Zareba W, Nomura A, Couderc JP. Cardiovascular effects of air pollution: What to measure in ECG? Environ Health Perspect 2001; 109 (suppl 4): 533-538.
Pulse Wave Analyses and Arterial Pulse Wave Velocity (SphygmoCor)
The basis of using the SphygmoCor device ( is that it provides a non-invasive measure of central aortic blood pressures (BP). These aortic BPs differ from arm levels depending on a variety of clinical parameters and are superior predictors of cardiovascular-related target organ status/damage and future cardiovascular events1,2. The device measures aortic augmentation index (an index of arterial pressure wave reflection from the periphery back to the aorta and overall left ventricular load) and a coronary perfusion index. These parameters are provided by pulse wave analyses of the radial artery using applanation tonometry and a mathematical generalized transfer function to determine the central arterial hemodynamics. The device also provides the gold-standard measurement of arterial stiffness by measurement of pulse wave velocity. These results have independent cardiovascular prognostic abilities superior to and incremental to standard arm BPs and are better measures of the hemodynamics imposed upon the heart, brain, and kidneys are thus superior predictors of target organ damage/status1,2.
In brief, after resting supine for 10 minutes, dominant arm radial artery applanation tonometry was performed for 10 seconds per the operational guidelines of the device for assessment of pulse wave analyses. Afterward, applanation tonometry was measured at the right carotid and femoral artery for 10 seconds each. Three-lead ECG recordings were obtained simultaneously with the tonometry in order to calculate pulse wave velocity. Semi-automated results were provided by the internal calculations of the device (i.e. generalized transfer function) to provide 1) Central aortic systolic, diastolic, and pulse pressure 2) Aortic augmentation pressure (AP), augmentation index (AIx), and AIx standardized to heart rate of 75 beats per minute (AIx@75). These are indices of arterial wave reflection at the central aorta and the overall left ventricular load. Central blood pressures and AIx are independent cardiovascular risk predictors and indicators of underlying physiology and hemodynamics. 3) Ejection duration, subendocardial viability ratio (an index of coronary artery perfusion and possible ischemia risk) 4) Aortic pulse wave velocity (PWV) a direct measure of arterial stiffness (gold standard measure of aortic artery compliance).
References
1. Laurent S, Cockcroft J, Van Bortel L, et al. Expert consensus document on arterial stiffness: methodological issues and clinical applications. Eur Heart J 2006; 27: 2588-2605.
2. Agabiti-Rosei E, Mancia G, O’Rourke MF, et al. Central blood pressure measurements and antihypertensive therapy. A Consensus Document. Hypertension 2007; 50: 154-160.
Microvascular Endothelial Function (EndoPat2000)
The endoPAT2000 device ( is FDA-approved to measure finger microvascular endothelial function in a fully-automated fashion. The device uses finger pulse amplitude tonometry (PAT) and measures PAT before (baseline) versus after reactive hyperemia (RH), termed RH-PAT, induced by upper-arm cuff occlusion for 5 minutes1-4. The major portion (60%) of the magnitude of the increase in the RH-PAT response is determined to be endothelial-dependent vasodilatation due to nitric oxide release. Large epidemiological studies have shown that the RH-PAT response standardized to the non-occluded arm PAT (RH-PAT index) is independently predicted by traditional and novel cardiovascular risk factors, and thus a determinant of overall cardiovascular risk1-4. The main outcome of RH-PAT index represents microvascular endothelial-dependent vasodilatation. Though correlated with conduit artery endothelial function in the coronary and brachial arteries (e.g. brachial FMD), it is a different arteriolar level territory and thus provides complimentary data to brachial FMD.
In brief, 5 minutes of bilateral basal resting finger PAT was recorded in resting position in both hands on one finger by the EndoPat probes. During this baseline period (at minute 4), basal brachial artery diameter (BAD) was recorded by Duplex ultrasound in the same dominant arm for at least 10 seconds (8-10 images) to be used for brachial FMD measurement. A dominant upper arm BP cuff was next inflated to 50 mm Hg above systolic BP to occlude blood flow for 5 minutes. Upon rapid cuff deflation, reactive hyperemia was created. RH-PAT was then recorded in the ipsilateral dominant hand finger for 5 minutes post BP cuff release. The device’s computer compared 120 seconds of baseline mean PAT (during the initial 5 minute period on the dominant arm) to the RH-PAT, defined as the mean PAT from 60-120 seconds post cuff release on the same dominant arm. These readings were then standardized to the PAT of the contralateral hand during these same time periods to provide the RH-PAT index (RI). Concomitantly, at 50 seconds to 90 seconds post release of the BP cuff during the reactive hyperemia period, reactive hyperemia BAD was also continuously recorded by Duplex ultrasound in order to calculate brachial artery FMD (see brachial FMD).
At the end of the second post cuff deflation 5-minute period, the subject was given nitroglycerin 0.4 mg sublingual. Non-dominant arm finger PAT was recorded for an additional 6 minutes. The PAT level for a 1-minute period occurring between 5-6 minutes after the nitroglycerin was recorded and then compared to a 1-minute period in the same non-dominant arm during the original baseline period prior to RH-PAT (from minute 4-5). This provided nitroglycerin-mediated PAT index (NTG-PAT). NTG-PAT is a measurement of endothelial-independent smooth muscle-mediated dilatation and is used to compare to the RH-PAT index (RI) to determine if alterations in vascular reactivity occur due to changes in endothelial-dependent or independent (smooth muscle) induced dilatation (or both).
References
1. Celermaher DS. Reliable endothelial function testing. At our fingertips? Circulation 2008;117: 2428-30.
2. Nohria A, Gerhard-Herman M, Creager MA, et al. Role of nitric oxide in the regulation of digital pulse volume amplitude in humans. J Appl Physiol 2006; 101: 545-48.
3. Hamburg NM, Keyes MJ, Larson MG, et al. Cross-sectional relations of digital vascular function to cardiovascular risk factors in the Framingham Heart Study. Circulation 2008; 117: 2467-74.
4. Kuvin JT, Patel AR, Sliney KA, et al. Assessment of peripheral vascular endothelial function with finger arterial pulse wave amplitude. Am Heart J 2003; 146: 168-74.
1