Personal Black Carbon Exposure Influences Ambulatory Blood Pressure: Air Pollution and Cardio-metabolic Disease (AIRCMD-China) Study

Abstract

Few prospective studies have assessed the blood pressure impact of extremely high air pollution encountered in Asia’s megacities. The objective of this study was to evaluate the association between combustion-related air pollution with ambulatory blood pressure and autonomic function. During February to July 2012, personal black carbon was determined for 5 consecutive days using microaethelometers in patients with metabolic syndrome in Beijing, China. Simultaneous ambient fine particulate matter concentration was obtained from the Beijing Municipal Environmental Monitoring Center and the U.S. Embassy. 24-hour ambulatory blood pressure and heart rate variability were measured from Day 4. Arterial stiffness and endothelial function were obtained at the end of Day 5. For statistical analysis, we used generalized additive mixed models for repeated outcomes and generalized linear models for single/summary outcomes. Mean (standard deviation) of personal black carbon and fine particulate matter over 24-hour was 4.66 (2.89) and 64.2 (36.9) μg/m³. Exposure to high levels of black carbon in the preceding hours was significantly associated with adverse cardiovascular responses. A unit increase in personal black carbon over the previous 10 hours was associated with an increase in systolic blood pressure of 0.53 mmHg and diastolic blood pressure of 0.37 mmHg (95% confidence interval, 0.17-0.89 and 0.10-0.65 mmHg, respectively), a percent change in low frequency to high frequency ratio of 5.11 and mean inter-beat interval of −0.06 (95% confidence interval, 0.62 to 9.60 and −0.11 to −0.01, respectively). These findings highlight the public health impact of air pollution and the importance of reducing air pollution.

INTRODUCTION

The updated global burden of disease report has once again highlighted the importance of air pollution as an important risk factor contributing to global mortality.¹ Nearly 90% of the world lives in regions exceeding the World Health Organization Air Quality annual standards for fine particulate matter <2.5 μm (PM_2.5).² In East Asia, PM_2.5 ranks as the 4^th leading risk factor for premature death as its mega-cities face some of the highest concentrations in the world.^2,3 We and others have provided evidence that ambient air pollution exposure is associated with increases in blood pressure (BP), likely via acute autonomic imbalance, which together may represent a plausible mechanism of air pollution-mediated acute cardiovascular events.^4-6 These associations have typically been reported at relatively low ambient levels of PM_2.5 in North America and Europe. Whether these adverse hemodynamic and autonomic responses persist in relation to the >10-fold higher PM_2.5 concentrations encountered in East Asia is unknown.^7,8 This is important in light of analyses demonstrating that the dose-response relationship for mortality due to PM_2.5 is attenuated at higher concentrations.⁹ In this prospective study, we investigated the association between personal level exposure to black carbon (BC) as well as ambient PM_2.5 with 24-hour ambulatory blood pressure (ABP) and 24-hour heart rate variability (HRV) in a cohort of individuals with the metabolic syndrome living in Beijing, China and who are chronically exposed to high pollution levels.

METHODS

Study population and design

Subjects with metabolic syndrome (n=65) were recruited from clinics affiliated with the Peking Union Medical College (PUMC) Hospital. The main motivation to conduct our experiments in metabolic syndrome was to study a patient group at high risk for transitioning to overt Type II diabetes mellitus. The Institutional Review Board (IRB) at PUMC Hospital approved the protocol and every subject signed a written informed consent (NCT01548300). Eligibility criteria included non-smoking adults between 35-75 years living in a non-smoking home in Beijing. Metabolic syndrome was defined by International Diabetes Federation (IDF) criteria specific for Asians, waist circumference >90 cm in males and >80 cm in females plus any two of the following: triglyceride level >150 mg/dL, high-density lipoprotein <40 mg/dL in males and <50 mg/dL in females, systolic BP >130mmHg, fasting plasma glucose >100 mg/dL or previously diagnosed Type 2 diabetes mellitus. Exclusion criteria included, smoking within past one year, self-reported daily secondhand smoke exposure >1 hour, severe occupational exposure to pollutants, intake of drugs that may alter baseline insulin sensitivity, or endothelial function (e.g. anti-oxidants, multi-vitamins, folic acid, fish oil supplementation, L-arginine) and use of non-steroidal anti-inflammatory drugs. The study protocol has been detailed previously.¹⁰ Briefly, during day 0, subjects arrived at the clinic and underwent baseline measurements including body mass index (BMI), waist circumference, and waist-to-hip ratio and were fit with the BC monitor and a global positioning device (USGlobalSat DG-100 Datalogger). The participants were encouraged to partake in their usual daily activities and maintain a diary during the 5-day period during which they wore personal exposure monitors. On day 4, they wore the ABP and Holter monitor for 24 hours and completed a diary defining their activities and locations for each half-hour period. On day 5, the subjects received tests that include (1) Endothelial function assessment (2) Pulse wave analysis and (3) Blood draws. All patients were enrolled between February 14, 2012 and July 14, 2012.

Exposure Measurements and Meteorological Variables

Personal BC was measured every 5 minutes over a five-day period using a microaethalometer (AethLabs MicroAeth AE51). As a quality check, time series of 5-minute ambient BC concentrations was collected continuously from a dual wavelength aethalometer (a model AE-20 Aethalometer; Magee Scientific, CA) at a fixed site in PUMC Hospital since May 28, 2012. In addition, ambient PM_2.5 concentrations were determined continuously using a tapered element oscillating microbalance sampler (hourly from TEOM operated at 50°C) at a fixed monitoring site at Tsinghua University (northwest Beijing, between the 4^th and 5^th Ring Roads) since June 15, 2012; official Beijing Municipal Environmental Monitoring Center (BMEMC) network data of hourly PM_2.5 was obtained for the Chegongzhuang site situated in the North 2^th Ring Road; postings of hourly PM_2.5 readings on a Twitter feed by U.S. Embassy in Beijing located at Northeast 3^rd Ring Road were also collected. Meteorological measurements of ambient temperature and relative humidity were obtained from the China Meteorological Administration for the whole study period.

24-hour Ambulatory Blood Pressure (ABP) Monitoring

ABP assessment was performed by a portable, lightweight, noninvasive monitor with a self-inflating cuff (Spacelabs Healthcare ABP Monitors; 90207, WA, USA). Systolic blood pressure (SBP), diastolic blood pressure (DBP) and heart rate (HR) were measured repeatedly every 20 minutes during daytime (0600h to 2200h) and every 30 minutes at night time (2200h to 0600h) over a 24-hour period starting from 9am on Day 4. Averages of ABP measurements (i.e. SBP, DBP, and HR) during daytime, night time and 24-hour were also calculated for further analysis.

24-hour Ambulatory Electrocardiogram (Holter ECG)

HRV assessment was performed over a 24-hour period on Day 4 for each subject using the 3-lead SpaceLabs recorder and the Impresario Holter system (SpaceLabs Healthcare, WA, USA). Frequency and time domain parameters were analyzed every 20 minutes during daytime and every 30 minutes at night time to match with ABP measurements. Frequency domain analysis provided estimates of the spectrum density of R-R intervals within specific frequency bandwidths, including total power (TP) (0.01–1.00 Hz), low frequency (LF) (0.04–0.15 Hz), high frequency (HF) (0.15–0.40 Hz) and the ratio of LF to HF (LF/HF). Time domain analysis included: SDNN, standard deviation of all N-N intervals obtained over the five minutes; pNN50, proportion of the N-N intervals greater than 50 milliseconds; rMSSD, square root of the mean squared differences of successive N-N intervals; meanRR, mean of the R-R intervals. All HRV indices were transformed logarithmically to meet the normality assumption.

Endothelial Function, Pulse Wave Velocity and Central Aortic BP Measurement

With the patient in supine position and after 30 minutes of resting quietly, the pulse wave velocity of the carotid and femoral arteries was analyzed, estimating the delay with respect to the electrocardiogram wave. Endothelial function was measured by means of peripheral arterial tonometry (EndoPat2000, Itamar Medical, Israel). The RH-PAT index reflected the extent of reactive hyperemia. Pulse wave velocity and central aortic blood pressure were estimated with the SphygmoCor CP non-invasive device (AtCor Medical, Australia).

Blood Tests

Fasting blood samples were drawn from subjects on day 5 and routine laboratory evaluations were conducted (lipid profile, creatinine and glycated hemoglobin.

Statistical Analysis

Construction of different exposure measures from the raw data

Daily means of PM_2.5 at three monitoring sites across Beijing (Tsinghua University, BMEMC, U.S. Embassy) were calculated. The average of the daily means from three monitoring sites was used as a representation of integrated daily mean of ambient PM_2.5. Since there was no data on PM_2.5 at Tsinghua University before June 15^th, the integrated daily mean of ambient PM_2.5 in the BMEMC and U.S. Embassy sites prior to that date was used. Previous studies have suggested a sub-acute effect (2-day to 5-day averages) of PM_2.5 on blood pressure;^5,11 hence we calculated the averages for preceding 24 hours (1-day lag) and 48 to 72 hours (3-day lag) and averages of cumulative exposure to 3 and 5 days (3-day, 5-day averages) as illustrated in Figure S1, for each subject, before their clinical visit on day 5. Similarly, personal level BC concentrations were averaged at multiple lags (1-day, 3-day lags) and time intervals (3-day, 5-day averages) from 5-minute repeated recordings for each subject. To match with the repeated measures of ABP, 1-10 hours moving averages of personal level BC concentrations were also calculated, corresponding to each 20 or 30-minute interval of ABP and HRV measurement. Exposure variables averaged on multiple intervals approximately follow a normal distribution, and no transformation was applied.

Descriptive statistics

Summary level descriptive statistics were computed for demographic parameters, exposure characteristics, meteorological variables and functional end-points including 24-hour ABP, 24-hour HRV analysis, endothelial function, pulse wave velocity and blood test assessments. Pearson’s correlation coefficients were calculated to evaluate the relationships among daily means of exposure end-points, including ambient BC at PUMC, personal-level BC, and regional PM_2.5 from three monitoring sites.

Analysis of single/summary outcome measures

For outcomes that comprised of one value per subject, we used standard multiple linear regression models to examine their associations with air pollutants. In our models, we regressed averages of 24-hour HRV indices, averages of 24-hour ABP (daytime, night time or 24-hour average), endothelial function index or pulse wave velocity on personal BC or ambient PM_2.5 using different lagging (1-day, 3-day lags) or averaging intervals (3-day, 5-day averages), with age, gender and BMI adjusted. To capture potential non-linear effects of mean temperature and relative humidity (averaged over the 24 hours before the measurement of the outcome), we included two smoothing terms corresponding to these predictors under a generalized additive model. The significance of the smoothing term was then tested and the model was simplified to a linear model when appropriate, based on the decision of this test. Changes in ABP and percent changes in HRV associated with 1 μg/m³ increase in personal BC or a 10 μg/m³ increase in ambient PM_2.5 at different lagging or averaging time periods were estimated along with corresponding Wald-type confidence intervals. All models were evaluated for outliers and influential observations by residual diagnostics.

Analysis of outcome data with repeated measures

Because linear regression model does no account for within-subject correlation of repeated outcome measures, we also investigated the impact of personal level BC on the repeated measures of ABP or HRV indices using the generalized additive mixed model (GAMM). The temporal structure of the exposure-response relationship was assessed using 1 to 10-hour moving averages of BC concentrations before each ABP measurements. All models controlled for age, gender, BMI, hypertension (yes/no), and diabetes (yes/no). Linear terms of mean temperature and relative humidity during the 24-hour period were found to be adequate and adjusted in the model. Resting heart rate was also included as a likely confounder for outcomes representing HRV indices. A penalized cubic regression spline function of time with 8 degrees of freedom was used to capture non-linear fixed effects of time and an autoregressive covariance structure among residual errors within subjects were used. This model choice was governed by the Akaike information criterion (AIC). Effects associated with 1 μg/m³ increase in personal BC at each averaging time period were estimated. To assess the potential effect of HR on ABP, sensitivity analyses were conducted, adjusting for the time-varying variable HR in the GAMMs with ABP as outcome. We also conducted sensitivity analyses including a factor representing microaethalometer assignment as a random effect, due to the concern of differences in BC measurement among devices.

All analyses were performed with statistical package R version 2.15.2 (www.r-project.org). Statistical significance was assessed using a 2-sided Wald test at the significance level of 0.05.

RESULTS

Baseline characteristics

Table 1 provides baseline patient characteristics, which included patients who met IDF criteria for metabolic syndrome with a mean HbA1c of 6.16%. Figure 1 depicts the personal level BC, ambient BC and ambient PM_2.5 concentrations from the three sites over the study period. Strong correlations were found for daily means of PM_2.5 from three fixed sites located in different parts of the city (r=0.85-0.95, p<0.0001) and daily means of BC at personal and ambient levels (r=0.84, p<0.0001) (Table S1). Personal BC also correlated well with regional PM_2.5 measurements with Pearson's correlation coefficients of 0.73-0.78 (p<0.0001) as illustrated in Table S1, although the mean concentration of personal BC (5.08 μg/m³) was <10% of ambient PM_2.5 (74 μg/m³) during the same period. Table 2 provides the summary of exposure concentrations, temperature and relative humidity for the subjects. The mean of 5-day average of ambient PM_2.5 (68.9 μg/m³, SD=25.2 μg/m³) was more than 3 fold higher than the U.S. National Air Quality Standard (NAAQS) (<15 μg/m³). The mean of 5-day average of personal BC (4.77 μg/m³, SD=1.76 μg/m³) was substantially higher than the average urban BC concentrations in North America (0.2-1.9 μg/m³).¹²

Table 1.

Participant characteristics (N=65)

Characteristics	Mean (SD)
Age	61 (9)
Female	37 (57)*
Body mass index (kg/m2)	26.0 (2.7)
Annual household income ≤ 40,000 RMB	29 (45)*
Hypertension	54 (83)*
Diabetes	17 (26)*
High Cholesterol	51 (78)*
SBP (mmHg)	125 (12)
DBP (mmHg)	76 (8)
heart rate (beats/min)	70 (9)
LF (msec2)	382 (316)
HF (msec2)	310 (286)
LF/HF	2.33 (1.20)
TP (msec2)	1457 (925)
SDNN (msec)	60 (14)
rMSSD (msec)	31 (12)
pNN50 (%)	6.5 (6.4)
meanRR (msec)	847 (111)
Low-density lipoprotein (mmol/L)	2.69 (0.86)
High-density lipoprotein (mmol/L)	1.25 (0.29)
Triglyceride (mmol/L)	1.63 (0.92)
Total cholesterol (mmol/L)	4.69 (1.16)
Creatinine (μmol/L)	67.5 (18.2)
Glycated hemoglobin (%)	6.16 (1.01)
Pulse wave velocity	7.54 (1.73)
Augmentation index at heart rate of 75 beats/min	24.5 (8.5)
Reactive Hyperemia Index	2.19 (0.55)

^*Results are presented as number (%).

Figure 1.

Time-series pattern of BC and PM_2.5 over the study period.

Daily means of ambient PM_2.5 were measured at U.S. Embassy in Beijing, Chegongzhuang site (Beijing Municipal Environmental Monitoring Center, BMEMC), and Tsinghua University. Daily means of personal level BC were expressed as mean ± SD. Ambient BC was monitored continuously at PUMC.

Table 2.

Descriptive statistics of air pollutants and meteorological parameters

Exposure	N	1-day lag	3-day lag	3-day average	5-day average
Personal black carbon (μg/m3)	64	4.66 (2.89)	3.86 (2.03)	4.65 (2.02)	4.77 (1.76)
Ambient PM2.5 (μg/m3 )	65	64.2 (36.9)	62.5 (52.5)	64.1 (29.2)	68.9 (25.2)
Temperature (°C)	65	19.7 (6.2)	19.7 (5.5)	19.9 (5.8)	19.8 (5.9)
Relative humidity (%)	65	56.9 (22.2)	49.7 (24.0)	54.8 (20.2)	55.8 (17.3)

Results are presented as mean (SD).

Effect of PM_2.5 and BC on ABP

Figure 2 displays the temporal variation of ABP and personal BC over a 24-hour period. ABP measures displayed normal diurnal variation over the 24-hour study period with a decrease during sleep and a surge in the early morning hours. Daily BC concentration displayed a bimodal pattern, with peaks between 7:00 and 8:00 am and between 7:00 and 11:00 pm, and with the lowest levels generally appearing around noon.

Figure 2.

Means (95% CIs) of ABP measures and personal level BC concentration during the 24-hour period (9 am on Day 4 to 9 am on Day 5).

Night time (10 pm to 6 am) is highlighted in grey.

In GAMM analysis, we observed positive association between personal BC and blood pressure for averaging periods ranging from 6 to 10 hours (Figure 3). Effects of BC on changes for ambulatory SBP were slightly greater in magnitude than for DBP across all the time periods examined, with an increase of 0.53 mm Hg (95% CI 0.17 to 0.89 mmHg) in SBP and an increase of 0.37 mm Hg (95% CI 0.10 to 0.65 mmHg) in DBP for 1 μg/m³ increase in BC during the previous 10 hours. We observed that the 10-hour moving average resulted in the greatest changes of ambulatory SBP, while 9-hour moving average resulted in the greatest change of DBP (0.39 mmHg, 95% CI 0.12 to 0.66 mmHg) and heart rate (0.31 beats/min, 95% CI 0.07 to 0.55 beats/min). Inclusion of microaethalometer assignment as a random effect or adjusting for heart rate as a time varying covariate for outcomes SBP and DBP did not appreciably change the results.

Figure 3.

Estimated changes and their 95% CIs in SBP, DBP and HR associated with 1 μg/m³ increases in 1-10 hour moving averages of BC using generalized additive mixed models (N=62).

All models controlled for subject, age, gender, BMI, temperature, relative humidity, hypertension, diabetes, and time of day. MA-1h, 1-hour moving average of BC before each ABP measurement.

Personal levels of BC averaged over the lagging and averaging intervals on the scale of days depicted in Table S2 however did not show any significant association with aggregate summary measures of ABP (daytime average, night time average or 24-hour average). Similarly, no significant association between cumulative averages or single day lags for PM_2.5 exposure and summary measures of ABP was observed (Table S3).

Effect of PM_2.5 and BC on HRV indices

Multivariable-adjusted associations of HRV indices with 1-10 hour moving averages of personal level BC concentration using GAMM are presented in Figure 4. The 7-hour moving average of personal BC was found to be marginally associated with LF/HF and mean RR interval. Longer averaging times showed stronger magnitudes of association in LF/HF and mean RR interval, with the largest associations exhibited for the 10-hour moving average that a 1 μg/m³ increment in personal BC was associated with 5.11 (95%CI 0.62 to 9.60) percent increase of LF/HF and 0.06 (95%CI 0.01 to 0.11) percent reduction of mean RR interval, respectively. No evidence of statistically significant association between personal BC concentration and any other HRV indices was observed. Consistent results were found when adding the microaethalometer assignment as a random effect in the models.

Figure 4.

Estimated percent changes and their 95% CIs in HRV indices associated with 1 μg/m³ increases in 1-10 hour moving averages of BC using generalized additive mixed models (N=63).

All models controlled for subject, age, gender, BMI, resting heart rate, temperature, relative humidity, hypertension, diabetes, and time of day. MA-1h, 1-hour moving average of BC before each ABP measurement.

Associations of PM_2.5 and BC over multiple lagging and averaging intervals on the scale of days with summary measures of HRV indices were assessed. However, no significant change in 24-hour average of HRV indices was seen with exposure to day lags or averages of personal level BC (Table S4). Similarly, no change in HRV summary measures was associated with day lags or averages of PM_2.5 exposure (Table S5).

Effects of BC and PM_2.5 on endothelial function and arterial stiffness

There was no significant association between different lags or moving averages of BC or PM_2.5 with either endothelial function or arterial stiffness measures.

DISCUSSION

BC has gained attention as a product and marker of combustion-related anthropogenic air pollution, often showing independent health associations distinct from those induced by “background” ambient PM_2.5.¹³ Our results demonstrate an effect of anthropogenic combustion-related air pollution (BC) on ABP and HRV in patients with metabolic syndrome in Beijing, a city continually facing extreme levels of air pollutants.

Although prior associations between exposures to ambient air pollutants and BP have been reported, these generally have been observed to occur in response to concentrations that are several folds lower.⁴ It is not clear whether adverse CV actions of air pollutants persist at the markedly higher levels encountered in Asia. Further, few studies have evaluated the effects of air pollutants in patients with metabolic syndrome, a population at risk for development of Type II diabetes and hypertension. Previous findings suggest that obese individuals and those with the metabolic syndrome may be at a greater risk.¹⁴ The rise in the prevalence of the metabolic syndrome as well as the marked levels of air pollutants in Beijing offered a unique opportunity to investigate this relationship.^15-17 Exposure to BC measured at the personal level over the prior 10 hours was associated with elevations in both SBP and DBP, which are consistent to the effect on ABP in studies performed in North America (Table S6). In the study by Delfino et al. in patients with coronary artery disease, outdoor BC showed small but significant associations with SBP and DBP especially under longer exposures, with each μg/m³ increase in BC over 8 hours resulting in an increase of 0.36 mmHg in SBP.⁵ Our findings demonstrate similar degrees of BP elevation per 1 μg/m³ change in BC, despite the levels of exposure being approximately 3-fold higher. Other studies that have examined the associations with BC and BP involved sitting clinic BP and lacked the ability to resolve BP and BC measurements hourly.^18-20 Differences in study design and the use of sphygmomanometric measurement of sitting BP rather than ABP make direct comparison difficult. Two prior studies in Beijing have shown the impact of traffic related air pollution on BP and a beneficial effect of facemasks in reducing BP.^21,22 In the first study conducted in healthy volunteers, a 2-hour walk with a facemask was associated with a 7 mmHg lower SBP and improved 24-hour HRV measures.²¹ A subsequent study among subjects with stable coronary artery disease demonstrated a favorable effect of facemask intervention in reducing ABP (3 mmHg reduction) during the walk along with improvement in HRV indices.²² In controlled exposure studies, we and others have demonstrated an increase in BP of a similar magnitude within hours of exposure to PM_2.5.^23,24 The doses used in these studies were accomplished with concentrators, that while unrealistic in North America, simulate concentrations routinely encountered in China.⁸ Thus the consistency of our observations across a range of doses, study designs and importantly in a susceptible population, render our findings that much more relevant.

Several mechanisms for the acute effects of inhaled particulates and BP have been proposed. Activation of the sympathetic nervous system via reflex arcs originating from the lung, altered vascular tone secondary to rapid endothelial dysfunction or reduced arterial stiffness and impaired baroreceptor sensitivity.^3,4 Our data seem to suggest small effects of BC and autonomic tone within the same time frame as the effects on BP. The 7-hour moving average of personal BC was marginally associated with LF/HF and mean RR interval. The largest association was seen with the 10-hour moving average, with increments in personal BC associated with increase in LF/HF and reduction of mean RR intervals. No association was found between indices of endothelial function or arterial stiffness and exposure to BC, suggesting that at least at this level of exposure, changes in these pathways are unlikely to influence BP.

Contrary to consistent significant findings with personal BC, our analyses with ambient levels of PM_2.5 did not exhibit any association with cardiovascular outcomes. There could be several reasons responsible for the lack of significance with PM_2.5. First, ambient level PM_2.5 was used as a proxy for personal PM_2.5 exposure in the assessment of the exposure-response relationship in our study. There is evidence for a large gradation and even discordance in exposure doses between ambient measures of PM_2.5 and personal PM_2.5.²⁵ Previous studies have suggested that personal-level exposure to PM_2.5 may elicit different responses than background ambient levels due to the varying sources and/or chemical composition of PM_2.5.^26,27 A second reason may have to do with the temporal resolution of our measures. Growing evidences have shown that the time course of the PM_2.5 effect on HRV occurred acutely within 24 hours.^28,29 However, we assessed associations with lagged and cumulative PM_2.5 on the scale of days rather than hours, precluding further investigation of exposure-response relationship at a higher temporal resolution. Thirdly, it has been observed that the dose-response relationship for mortality due to PM_2.5 is not a linear function, with lower slopes at higher concentrations.⁹ Therefore, we explored the possibility of non-linear exposure effect on cardiovascular outcomes by including a smoothing term corresponding to 5-day average of ambient PM_2.5 concentration under a generalized additive model, with little statistical evidence of non-linearity (Figure S2). Finally, our study was conducted with a relatively small sample size and the PM_2.5 exposure measurements among patients examined in the same week at their first study visit may have contributed to limited variability due to the use of integrated daily mean of ambient PM_2.5, thereby reducing the power for detecting a significant exposure effect.

There were several strengths of the current study, including measurement of exposure and outcomes in metabolic syndrome patients who represent an “at-risk” patient population; the use of personal level BC monitoring and multiple functional outcomes to provide read-outs on putative mechanistic pathways. We acknowledge that although we used BC as a surrogate for anthropogenic sources of air pollution, the precise components of traffic related pollutants that contribute to BP remain elusive. One additional limitation is the lack of information on personal level PM_2.5 or other co-pollutants. Thus we cannot rule out the possibility that personal level PM_2.5 or other fractions of PM_2.5 may have demonstrated an association with BP. In general, prior studies have demonstrated that the correlation between outdoor level of traffic related pollutants such as BC and nitrogen-dioxide with PM_2.5 are strong.³⁰ Indeed, there was a high degree of correlation between personal BC and PM_2.5, while ambient BC closely mirrored personal BC. A final limitation may be that our findings may not be generalizable to all patients.^3,31,32

PERSPECTIVES

Our study demonstrates an important linkage between elevated BP and altered HRV and exposure to combustion-related pollutants. These findings highlight the public health impact of air pollution and the importance of reducing anthropogenic air pollution.

NOVELTY AND SIGNIFICANCE

What is new?

Few prospective studies have evaluated the association between personal measures of traffic related air pollution and blood pressure in susceptible populations, in environments associated with extremely high levels of air pollutants.

What is relevant?

In this study, we describe an association between personal BC (a measure of traffic related air pollution) and ambulatory SBP and DBP. Changes in BC were also associated with alterations in HRV indices.

Summary

Our findings suggest that the relationship between urbanization and hypertension may have complex underpinnings, with factors such as air pollution contributing to increases in BP via autonomic alterations.