Short-term exposure to traffic-related emissions was associated with significant acute changes in HRV in this panel study of researchers that participated in the Mexico City Air Pollution Campaign . Gaseous pollutants – particularly ozone – were associated with reductions in time and frequency domain components. In contrast, PM2.5 increased these HRV parameters. Like Riediker , our results show a positive association between PM2.5 (mean PM2.5 mass concentration = 23.0 μg/m3) and two frequency domain HRV parameters (HF and LF), a result contrary to those observed in multiple studies of elderly populations exposed to PM air pollution related to traffic [13, 43, 44] (median PM2.5 mass concentration in each study, respectively = 10 μg/m3, 8.92 μg/m3 (2-hr), 7.7 μg/m3 (5 minute). Our results also demonstrate a negative association between the LF/HF ratio and PM2.5, CO, and CO2 that is significant for the 30-min averaging period. When assessed in normalized units (i.e. ratio), LF and HF provide quantitative indicators of neural control of the sinoatrial node and provide a synthetic index of the sympathovagal balance , which may be predictive of the development of ventricular arrhythmias . Our results, similar to other studies discussed below, suggest that the air pollution-associated acute changes in HRV parameters (increased or reduced) that have the highest potential for increasing the likelihood of a subsequent cardiac arrhythmia remain to be determined. Nonetheless, our results have significant implications to our understanding of how air pollution leads to an increase risk of cardiac arrhythmias. While having a reduced HRV is a risk factor for increased cardiovascular mortality; it is yet to be shown whether short-term environmental exposures associated with acute HRV changes can lead to cardiac arrhythmias in humans. This pathway, however, seems biologically plausible given that a) acute reductions or increases in HRV parameters have been associated with the onset of ventricular tachycardia [47–49], and b) short-term increases in ambient pollutants (both particles and gases) have been associated with increased likelihood of having a discharge from an implantable cardiac defibrillator .
Wu et al.  measured real-time, in-vehicle, traffic-related PM2.5 (56.6 μg/m3 (daily average)), and gaseous co-pollutants (CO, NO2, and NO) in a young population (n = 11, mean age = 35.5 years). They showed that IQR increases in PM2.5 mass concentrations (5–240 min moving averages) were associated with declines in three, 5-min HRV indices (SDNN, LF, and HF). Results, however, from their regression models for each subject showed heterogeneity among responses, i.e. several subjects in the study had positive associations with traffic-related PM exposures for the three, 5-min HRV indices. Further, their smoothed curves showing associations between PM exposure and 5-min HRV indices indicated that lower PM exposures were associated with increases in HRV, whereas higher PM exposures were associated with decreases in HRV. Similarly, to explore potential confounding by the gaseous co-pollutants, we separately included CO, NOx, and NO in a two-pollutant model with PM2.5. While adjusting for the co-pollutants decreased the precision of the estimates, our overall results were generally consistent with estimates not adjusted for co-pollutants. Overall, the results of Wu et al., like ours, suggest that differences in HRV response to traffic-related PM pollution may be related to differences in exposure levels, though factors impacting the heterogeneity of responses remain unclear.
Other studies in young populations illustrate the heterogeneity of responses between PM exposure and HRV. In a study of mail carriers between 25 and 46 years of age in Taiwan, no significant HRV effects from PM (mean sample time of approximately four hours using a personal cascade impactor sampler with a pump) were observed despite reported higher traffic-related air pollution exposure levels as compared to other studies (median PM2.5 exposure = 61.3 μg/m3) . Similarly, PM2.5 exposure assessed by a light-scattering method (pDR real time instrument, 0–180 min averaging time) did not affect the SDNN measurements in 40 young, healthy residents of the Mexican metropolitan area . In a population of nine young highway patrol officers (mean age = 27.3 years), Riediker et al.  showed that in-vehicle exposure to PM2.5 (measured both by gravimetric and light-scattering methods) was associated with all time domain parameters (% difference between adjacent normal RR intervals that are greater than 50 msec or PNN50, SDNN, and mean cycle length, a phrase to emphasize that the interval between consecutive beats, rather than the heart rate, is being analyzed), as well as HF power and the power ratio LF/HF on the morning after the shift. In the occupational literature, Magari et al.  reported statistically significant associations using a two-hour lagged mean heart rate. Specifically, the PM2.5 average air concentration (measured using a light scattering instrument, mean 1160 μg/m3) showed an average increase of 1.67 msec (95% CI: 0.11 to 3.22) in the mean heart rate, for every 1 mg/m3 increase in the average PM2.5 concentration. The variety of exposure assessment techniques and corresponding averaging times may also contribute to the perceived heterogeneity of responses.
In addition to different PM exposure metrics, studies have used various methods to estimate the impact of gaseous co-pollutants. Some have used fixed-site monitors [13, 20, 30, 43] and/or time activity data  to estimate traffic-related ambient air pollution exposure, including gaseous co-pollutants. Our results showing acute reductions in HRV in association with ozone and other gaseous pollutants is in agreement with other recent studies .
In a repeated measures study of 46 subjects (43–75 years of age), Zanobetti et al.  suggested that pollutant mixtures may influence cardiac tone as both PM and O3 (30–120 min moving averages) had independent associations with reduced HRV in two-pollutant models. Using a five minute time resolution to evaluate the acute effects of residential outdoor ozone exposure and HRV changes, Jia et al.  showed in 20 elderly subjects that, after adjusting for other pollutants and subject characteristics, there was a reduction in the HF component of 4.87% (95% CI:0.97 to 8.62) per 10 ppb increment of O3. A similar result was observed in a study of patients recently discharged from the hospital for acute coronary disease-related complications. Here, the two-hour and five-day O3 moving averages were associated with reductions in time domain components of HRV indicative of parasympathetic function, whereas NO2 was associated with reductions in the HF spectral component .
A major limitation of these studies (as referenced in both) was the use of fixed-site monitors to estimate personal exposure measurements. Suh and Zanobetti  showed that changes in HRV – especially those associated with parasympathetic control – were significantly and negatively associated with elemental carbon, and, to a lesser degree, NOx when measurements of personal exposure (but not ambient, outdoor or indoor concentrations) were used to estimate their exposures. Interestingly and importantly, associations between personal exposure measurements and HRV were detectable only for these traffic-related pollutants. Non-significant findings with HRV were detected for 24-hour ambient concentrations and personal exposures to more spatially uniform regional pollutants, i.e. PM2.5 and O3.
Similar results were reported in a large controlled-exposure study examining the association between HRV and combined exposure to concentrated ambient particles (CAPs) and O3. In the participating group of healthy young adults (n = 50, mean age = 27.1 years), no consistent pattern of changes in HRV indices was detected among PM2.5 and O3 exposure categories . Despite the absence of a clear trend in the categorical exposure models, the dose–response analysis demonstrated a trend toward a negative linear association between CAPs mass concentration and change in several HRV indices, with a statically significant relationship for the LF HRV measure. No relationship existed without accounting for O3.
Given the similarities between these studies with our results, it is possible to speculate that, regardless of age or underlying cardiovascular disease, acute ozone exposures are associated with reductions in HRV. With respect to acute PM2.5 exposures, there are notable age-dependent and/or possibly underlying cardiovascular diseases that may influence the HRV responses to PM. In addition, acute changes in HRV (even those characterized by an increase in the HF domain) can be associated with the onset of a ventricular arrhythmia .
To the best of our knowledge, the literature contains varied to little information on the relationship between HRV parameters and potential associations between NOx, CO2, and formaldehyde. The selection of exposure measures reported here was not based on an a priori reason that some evidence existed to support association between exposure and HRV. Rather we evaluated potential associations within available exposures as part of the Mexico City Air Pollution Campaign. In 2004, the Multi-Ethnic Study of Atherosclerosis and Air Pollution was launched to investigate the relation between individual-level estimates of long-term air pollution exposure and the progression of subclinical atherosclerosis and the incidence of cardiovascular disease and includes exposure assessments of fine particulate matter, NOx, and black carbon; the majority of data collection will be completed in 2014 . Some have shown that certain nitrogen oxides, e.g. NO2, do not affect heart rate variability at concentrations high for urban background levels and in the absence of other pollutants . Others have shown that ambient NO2 concentrations were inversely associated with SDNN and positively associated with LF/HF. (β = 1.4; 95% CI, 0.35 to 2.5) 2 hr after the start of cycling . Our results demonstrate that both NO2 and NOx are significantly associated with SDNN. No studies were located that demonstrated a cardiovascular effects in humans after inhalation exposure to CO2 or formaldehyde . In our case, it’s likely that CO2 and/or formaldehyde exposures were acting as a surrogate for other vehicular gaseous pollutants.
One limitation of this work is that this analysis did not control for stress resulting from traffic. Psychological stress can influence both HRV and autonomic function . Since we did not measure participant stress while in traffic, we cannot disentangle the impact of both pollution and stress exposures resulting from being in traffic. In addition to traffic-related stress, participants may have observed the measured pollutant concentrations while being transported in the van-based mobile laboratory. Knowledge of their exposures may have stimulated the participants' sympathetic tone. Secondly, any exposure misclassification can lead to biased estimates of exposure-response estimates, particularly in cases with multiple correlated exposures where the direction of the bias is uncertain . While exposure error biases are often assumed to bias toward the null, a more complicated situation arises in cases like ours when two or more exposures are measured with error and are correlated with each other. This may lead to bias in both directions and with varying degree. A final caveat is that chemical exposure factors are known to affect HRV [61, 62], but these exposures are unlikely to be present in our study and thus, would be unlikely to affect our findings.