Concentration-Response Functions Relating Concentration and Duration of Ambient Traffic-Related Air Pollution Exposure to Systolic Blood Pressure

Abstract

Eliasziw et al discuss their study on the short-term concentration--response data from a randomized crossover trial examining differences in blood pressure (BP) associated with exposure to ultrafine particles (UFP) and black carbon (BC). The trial results showed that portable high-efficiency particulate arrestance (HEPA) filters reduced indoor infiltration of TRAP and were effective at preventing short-term increases in systolic blood pressure (SBP). The trial data are amenable to deriving concentration-response functions that incorporate both concentration and duration of UFP and BC exposure.

Introduction

Exposure to traffic-related air pollution (TRAP) is associated with adverse health outcomes, including elevated blood pressure (BP).1-3 However, unlike fine particulate matter [PM =2:5 lm in aerodynamic diameter (PM2:5)] for which concentration-response functions have been characterized for chronic4 and acute5 effects, there are few reports in the literature describing associations for ultrafine particles (UFP) and black carbon (BC).6,7

The present study analyzed the short-term concentration- response data from a randomized crossover trial examining dif-ferences in BP associated with exposure to UFP and BC. The trial results showed that portable high-efficiency particulate arrestance (HEPA) filters reduced indoor infiltration of TRAP and were effective at preventing short-term increases in systolic blood pressure (SBP).8 The trial data are amenable to deriving concentration-response functions that incorporate both concen-tration and duration of UFP and BC exposure, which we present here.

Methods

Details of the randomized, three-exposure, three-period crossover trial have been published.8 Adults 40-75 years of age spent 2 h during morning rush hour in rooms near urban highways on three consecutive weeks, separated by 1-wk washout periods. During each weekly session, participants sat quietly wearing sound can-celing headphones. BP was measured every 10 min with an am-bulatory monitor (SunTech Medical; Oscar) on the dominant arm. All participants provided informed consent and the Tufts University Social Sciences Institutional Review Board approved the study protocol.

Low, medium, and high TRAP exposure conditions were cre-ated by varying the degree of ventilation (and thereby infiltration) and by moderating the rate of air flow through the HEPA filtration units. We sought a 10-fold concentration contrast between high-and low-exposure sessions; the mean particle number and BC concentrations were respectively 30,000 particles=cm3 and 830ng=m3 during high exposure, and 2,500 particles=cm3 and 150 ng=m3, during low exposure. For medium-exposure ses-sions, we aimed for 10,000 particles=cm3; the mean particle number and BC concentrations were 11,000particles=cm3 and 410ng=m3. To the extent possible, in a real-world exposure scenario, every effort was made to blind participants to the exposure level. Particle number concentration (PNC, which primarily meas-ures UFP because larger PM is present in much smaller numbers) was measured continuously using a water-based condensation particle counter (TSI, Inc; model 3873; d50 = 7nm) at 1-s resolution. BC concentration was measured with an aethalometer (Magee Scientific; model AE16) at 1-min resolution. For the purposes of analysis, 10-min averages of PNC and BC concentrations were cal-culated to correspond to the 10-min intervals between SBP measure-ments. Measurements were excluded from the first 20 min of exposure because we observed a drop in SBP within the first 20 min of the start of each exposure period. The final half hour of the exposure sessions was also excluded from the analyses because the data were deemed to be unreliable as participants became restless and restaurant cooking odors became apparent in some sessions.

Regression models using generalized estimation equations were used to fit logarithmic relationships between PNC and SBP and between BC and SBP. Concentrations were natural log trans-formed based on empirical evidence that the shape of the three-dimensional concentration-response surface is nonlinear.4 The models included the variable time from participant entry into the exposure room as a measure of exposure duration. Robust standard errors were used in the inferences about the estimated regression coefficients. All statistical analyses were carried out using SAS (version 9.4; SAS Institute Inc.), and results with p<0:05 were deemed statistically significant.

Results

A total of 77 participants were analyzed. Their average age was 60 y, 61 (79%) were female, 59 (77%) were of Chinese ethnicity, 13 (17%) were White, and 5 (6%) were Black or Hispanic. Complete demo-graphics are provided in Hudda et al.8 None reported serious health conditions. The temperature of the room remained relatively constant throughout each session (average coefficient of variation = 1:4%) at 22° C, on average. Each of the 77 participants contributed between 13 and 21 SBP measurements, with corresponding measurements of PNC and BC concentrations, for a total 1,470 measurements. Although the maximum sample size would be expected to be 1,617 measurements (77 participants × 7 BP measurements × 3 periods), because of missing BP, PNC, and BC data, 1,470 (91%) measurements were available to be analyzed. The concentrations during the controlled exposure sessions were relatively stable over time but varied in response to fluctuation in ambient conditions, resulting in an overlap in the range of concentrations measured during the low, medium, and high exposures. The data, differentiated by the three exposure conditions, are dis-played in Figure 1A,B.

Our models, which yielded statistically significant estimated regression coefficients, are displayed in Figure 2A,B and can be writ-ten as follows: SBP = 102:58+0:91 ln(PNC)+2:27 ln(DurationPNC) and SBP = 105:26+1:01 ln(BC)+2:21 ln(DurationBC). In these equations, ln(PNC), ln(BC), ln(DurationPNC), and ln(DurationBC) are the natural logarithms of PNC, BC concentration, and ex-posure duration. p-Values corresponding to the estimated regression coefficients were 0.027 for ln(PNC), 0.038 for ln (DurationPNC), 0.037 for ln(BC), and 0.043 for ln(DurationBC).

The similarity in regression coefficients between the PNC and BC models can be explained by the relatively high correlation between their concentrations (Pearson correlation = 0:79). Although both models describe a progressive dampening of the increase in SBP with increasing concentration and dura-tions of exposure, duration of exposure had a greater effect on SBP than did concentration. Whereas a 54% increase in dura-tion resulted in a 1-mmHg increase in SBP, the same increase in SBP required a 152% increase in concentration.

Discussion

The results suggest concentration-response functions for PNC and BC with SBP that are logarithmically dependent on the two dimensions of exposure: concentration and duration. Our find-ing is qualitatively similar to a prior model suggested for PM2:5, albeit the temporal and concentration scales were quite differ-ent.4 We propose it is reasonable that both concentration and du-ration contribute to biological effects given that both are determinants of exposure. Consideration of the concentration- response functions should inform future epidemiological studies of UFP and BC on BP.

The present study has both strengths and limitations. The continuous monitoring of PNC and BC concentrations, as well as the frequent intervals for BP measurements, yielded a large number of observations under well-controlled conditions. Frequent BP measurements on the same participant permitted an assessment of the effects of exposure duration. However, we did not measure other pollutants, notably PM2:5 and PM =10 lm in aerodynamic diameter (PM10), that could potentially confound our findings. Another limitation was that participants were primarily of Chinese ethnicity, for whom reduced responses to UFP have been reported,9 limiting generalizability. In addition, the concentration-response functions derived from adults 40-75 years of age may not be applicable to younger adults. Finally, the rela-tively short exposure duration restricts extrapolation to longer, real-life exposure scenarios. Nevertheless, given the paucity of data about concentration-response functions for near-roadway TRAP, our findings are a starting point for further investigation into these concentration-response functions.

Acknowledgments

This research was supported by the National Institute of Environmental Health Sciences (R01-ES026980 and R01-ES030289).

References

References

1. Mills NL, Miller MR, Lucking AJ, Beveridge J, Flint L, Boere AJ, et al. 2011. Combustion-derived nanoparticulate induces the adverse vascular effects of diesel exhaust inhalation. Eur Heart J 32(21):2660-2671, PMID: 21753226, https://doi.org/10. 1093/eurheartj/ehr195.

2. Peters A, von Klot S, Mittleman MA, Meisinger C, Hörmann A, Kuch B, et al. 2012. Triggering of acute myocardial infarction by different means of transportation. Eur J Prev Cardiol 20(5):750-758, PMID: 22544548, https://doi.org/10.1177/ 2047487312446672.

3. Matz CJ, Egyed M, Hocking R, Seenundun S, Charman N, Edmonds N. 2019. Human health effects of traffic-related air pollution (TRAP): a scoping review protocol. Syst Rev 8(1):223, PMID: 31464631, https://doi.org/10.1186/s13643-019-1106-5.

4. Pope CA III, Brook RD, Burnett RT, Dockery DW. 2011. How is cardiovascular disease mortality risk affected by duration and intensity of fine particulate matter exposure? An integration of the epidemiologic evidence. Air Qual Atmos Health 4(1):5-14, https://doi.org/10.1007/s11869-010-0082-7.

5. Orach J, Rider CF, Carlsten C. 2021. Concentration-dependent health effects of air pollution in controlled human exposures. Environ Int 150:106424, PMID: 33596522, https://doi.org/10.1016/j.envint.2021.106424.

6. Fakhri AA, Ilic LM, Wellenius GA, Urch B, Silverman F, Gold DR, et al. 2009. Autonomic effects of controlled fine particulate exposure in young healthy adults: effect modification by ozone. Environ Health Perspect 117(8):1287-1292, PMID: 19672410, https://doi.org/10.1289/ehp.0900541.

7. Magalhaes S, Baumgartner J, Weichenthal S. 2018. Impacts of exposure to black carbon, elemental carbon, and ultrafine particles from indoor and outdoor sour-ces on blood pressure in adults: a review of epidemiological evidence. Environ Res 161:345-353, PMID: 29195183, https://doi.org/10.1016/j.envres.2017.11.030.

8. Hudda N, Eliasziw M, Hersey SO, Reisner E, Brook RD, Zamore W, et al. 2021. Effect of reducing ambient traffic-related air pollution on blood pressure: a random-ized crossover trial. Hypertension 77(3):823-832, PMID: 33486990, https://doi.org/10. 1161/HYPERTENSIONAHA.120.15580.

9. Lane KJ, Levy JI, Scammell MK, Peters JL, Patton AP, Reisner E, et al. 2016. Association of modeled long-term personal exposure to ultrafine particles with inflammatory and coagulation biomarkers. Environ Int 92-93:173-182, PMID: 27107222, https://doi.org/10.1016/j.envint.2016.03.013.