Temperature, traffic-related air pollution, and heart rate variability in a panel of healthy adults

Environmental Research 120 (2013) 82–89

Contents lists available at SciVerse ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Temperature, traffic-related air pollution, and heart rate variability in a panel of healthy adults Shaowei Wu a,1, Furong Deng a,1, Youcheng Liu b, Masayuki Shima c, Jie Niu d, Qinsheng Huang d, Xinbiao Guo a,n a

Department of Occupational and Environmental Health Sciences, Peking University, School of Public Health, Beijing, China Department of Environmental and Occupational Health Sciences, School of Public Health, University of North Texas Health Science Center, Fort Worth, TX, USA c Department of Public Health, Hyogo College of Medicine, Hyogo, Japan d Peking University Third Hospital, Beijing, China b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 April 2012 Received in revised form 24 August 2012 Accepted 25 August 2012 Available online 18 September 2012

Background: Both ambient temperature and air pollution have been associated with alterations in cardiac autonomic function, but the responsive patterns associated with temperature exposure and the interactive effects of temperature and air pollution remain largely unclear. Objectives: We investigated the associations between personal temperature exposure and cardiac autonomic function as reflected by heart rate variability (HRV) in a panel of 14 healthy taxi drivers in the context of traffic-related air pollution. Methods: We collected real-time data on study subjects’ in-car exposures to temperature and trafficrelated air pollutants including particulate matter with an aerodynamic diameter r2.5 mm (PM2.5) and carbon monoxide (CO) and HRV indices during work time (8:30–21:00) on 48 sampling days in the warm season (May–September) and cold season (October–March). We applied mixed-effects models and loess models adjusting for potential confounders to examine the associations between temperature and HRV indices. Results: We found nonlinear relationships between temperature and HRV indices in both the warm and cold seasons. Linear regression stratified by temperature levels showed that increasing temperature levels were associated with declines in standard deviation of normal-to-normal intervals over different temperature strata and increases in low-frequency power and low-frequency:high-frequency ratio in higher temperature range (4 25 1C). PM2.5 and CO modified these associations to various extents. Conclusions: Temperature was associated with alterations in cardiac autonomic function in healthy adults in the context of traffic-related air pollution. & 2012 Elsevier Inc. All rights reserved.

Keywords: Air pollution Cardiac autonomic function Cardiovascular disease Heart rate variability Temperature

1. Introduction Significant changes in ambient temperature have been found to have adverse effects on human health as well as the air pollution (Kan et al., 2012), and studies during recent years have associated changes in both ambient temperature and air pollution with increased morbidity and mortality of cardiovascular disease (Goldberg et al., 2006; Guo et al., 2011; Ito et al., 2011; Ren et al., 2006; Stafoggia et al., 2008; Zanobetti and Schwartz,

Abbreviations: CO, Carbon monoxide; HRV, Heart rate variability; PM2.5, Particulate matter with an aerodynamic diameter r2.5 mm n Correspondence to. Department of Occupational and Environmental Health Sciences, Peking University School of Public Health, No. 38 Xueyuan Road, Beijing 100191, China. Fax: þ 86 10 62375580. E-mail address: [email protected] (X. Guo). 1 These authors contributed equally to the study. 0013-9351/$ - see front matter & 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.envres.2012.08.008

2008). Several biological mechanisms linking environmental exposures to the development of cardiovascular disease have been proposed with alterations in cardiac autonomic function as one of the mechanisms that received much attention (Hampel et al., 2012; He et al., 2011; Jia et al., 2011, 2012; Ren et al., 2011; Riediker et al., 2004; Wu et al., 2010; Zanobetti et al., 2010). Heart rate variability (HRV) is a noninvasive and sensitive measure of cardiac autonomic function and has been frequently used in previous studies to assess the cardiac effects of environmental exposures. However, compared with the robust literature focusing on the cardiac autonomic dysfunction associated with air pollution exposure, few studies have examined this mechanism in light of ambient temperature changes, and most of them have been conducted under experimental conditions (Bruce-Low et al., 2006; Sollers et al., 2002; Yamamoto et al., 2007; Yao et al., 2009). Previous epidemiologic studies have examined the cardiovascular effects of ambient temperature under various settings (Adamopoulos et al., 2010; Brook et al., 2011;

S. Wu et al. / Environmental Research 120 (2013) 82–89

Pope et al., 2004a; Ren et al., 2006, 2011; Stafoggia et al., 2008; Zanobetti and Schwartz, 2008), whereas only one of them has reported the association of ambient temperature with HRV as well as modification of this association by air pollution in an aging population (Ren et al., 2011). This study found that higher ambient temperature was associated with decreased HRV in the aging population in the warm season but not in the cold season. In fact, ambient temperature in the cold season would be a poor surrogate for real temperature exposure because people spend most time inside home and use heating in the cold season, and this inconsistency would conceivably conceal the association of ambient temperature with health outcomes. To date, the change patterns of human cardiac autonomic function in response to simultaneous exposures to temperature and air pollution and their interactive effects still remain largely unclear. Recently we have reported the association of personal exposure to traffic-related air pollution with HRV in a panel of healthy taxi drivers around the Beijing 2008 Olympic time, a period that underwent significant air quality changes due to the implementation of air pollution control measures by the government (Wu et al., 2010, 2011a, 2011b). In addition to the air pollution changes, we also observed significant ambient temperature changes over the study with the highest daily mean air temperature of 25.0 1C during the Olympics and the lowest daily mean air temperature of 2.6 1C during the winter period (heating season) (Wu et al., 2011a). The real-time, personal exposure measurements conducted in our study provide better estimation for subjects’ real exposures to temperature and air pollutants especially in the cold season. These give us the excellent opportunity to investigate the relationship between temperature exposure and HRV as well as the interactive effects of temperature and air pollution exposures. In the present analysis, we examined the effects of personal temperature exposure on HRV in the context of air pollution changes.

2. Methods 2.1. Study design and subjects We recruited 14 eligible subjects from 44 voluntary taxi drivers based on the following criteria: ageo 45 years, nonsmoking, no history of physician-diagnosed cardiovascular, pulmonary, neurological or endocrine diseases, body mass index r 30, drive taxicab during daytime hours, employed as a taxi driver for at least one year, and normal physical and blood examination results. We collected the following personal information using a self-administered questionnaire: name, sex, age, years of employment as a taxi driver, smoking status, education, and history of cardiovascular diseases or other diseases. Taxi drivers were then physically examined and tested for their seated blood pressures, resting electrocardiogram, blood cholesterol, triglycerides, and high- and low-density lipoproteins. Eligible subjects were followed for four time periods around the Beijing 2008 Olympic Games, which were before (late May to mid June, 2008), during (mid August to early September, 2008) and after (late October to mid November, 2008) the Beijing 2008 Olympic Games, and a subsequent winter period (late February to mid March, 2009), respectively. Each driver was measured for his/her ambulatory electrocardiogram and real-time, in-car exposures to temperature, relative humidity and two major traffic-related air pollutants (PM2.5 and CO) using specific instruments and materials for a separate daily work shift (08:30–21:00) during each of the four time periods on weekdays (Monday to Friday). The study was approved by the Institutional Review Board of Peking University Health Science Center, and informed consent was obtained from each subject before the study began.

2.2. Exposure data Instruments and materials for exposure measurements were placed on the front passenger seat parallel to the driver, and the sampling inlets or sensors were about 35–50 cm above the seat base. The drivers were asked to keep the status of taxi windows of both sides consistent with each other (open/closed) throughout the measurement periods for the equilibrium of air flow. Therefore, the measurement results could well represent the driver’s real exposure in personal levels. In

83

particular, real-time temperature and relative humidity were measured by a HOBO Pro V2 temperature/relative humidity logger (Onset Corp., Pocasset, MA, USA). Real-time PM2.5 concentrations were measured by a Grimm Model 1.109 Portable Aerosol Spectrometer (Grimm Technologies Incorporation, Douglasville, GA, USA), and real-time CO concentrations were measured by a Model T15n Enhanced CO Measurer (Langan Products Inc., San Francisco, CA, USA). Instruments used in the study were all calibrated or checked for measurement accuracy according to the manufacture’s specifications before the measurement began. The exposure measurements commonly began about half-hour (at 08:30) before the commencement of the electrocardiogram measurements, and all real-time exposure variables were logged in 1-min intervals and aggregated as 5-min averages. During the measurement periods, traffic routes of the taxicabs were recorded using a global positioning system device and the drivers were asked to record their personal activities in work diaries. Both drivers and passengers were not allowed to smoke inside the taxicab throughout the measurement periods.

2.3. Heart rate variability data Subjects were fitted with a standard 5-lead ambulatory electrocardiogram recorder (model MGY-H7; DM Software Inc., Stateline, NV, USA). Each subject’s skin was carefully cleaned with alcohol pads by a trained technician before the electrodes were connected. Each subject wore the recorder throughout the daily work shift (09:00 to 21:00) along with the exposure measurements. Data obtained from the recorder were processed using specific software (Holter System Ver 12.net, DM Software Inc., Stateline, NV, USA) according to standard criteria (Task Force, 1996). In the present analysis, we used the following four representative HRV indices which were commonly used to examine the effects of environmental exposures on cardiac autonomic function in previous studies (Jia et al., 2012; Hampel et al., 2012; He et al., 2011; Ren et al., 2011; Wu et al., 2010; Zanobetti et al., 2010): (a) standard deviation of normal-to-normal intervals; (b) lowfrequency power (0.04 to 0.15 Hz); (c) high-frequency power (0.15 to 0.40 Hz); and (d) low-frequency:high-frequency ratio: the ratio of low-frequency power to high-frequency power. The standard deviation of normal to normal intervals is an index in time-domain which reflects all the cyclic components responsible for variability in the period of recording, and the other three HRV indices lowfrequency power, high-frequency power and low-frequency:high-frequency ratio are indices in frequency-domain (Task Force, 1996). Among the frequency-domain indices, low-frequency power is an index reflecting both sympathetic and parasympathetic (vagal) influences whereas high-frequency power is an index reflecting parasympathetic tone, and low-frequency:high-frequency ratio is a method of assessing sympathetic and parasympathetic balance for the cardiac autonomic function (Task Force, 1996; Zanobetti et al., 2010). Previous clinical studies have demonstrated that these HRV indices were strong predictors of mortality in cardiac patients (La Rovere et al., 2003; Nolan et al., 1998), and air pollution exposure may induce increased cardiovascular mortality through the cardiac autonomic dysfunction pathway as reflected by changes in HRV (Pope et al., 2004b). In the present study, all the HRV indices were calculated in standard 5-min segments throughout the measurement periods. Data when drivers were out of their taxicabs were identified according to their work diaries and global positioning system records, and were excluded before analysis.

2.4. Statistical analysis The overall database was stratified by a warm season (with data from periods of before and during the Olympics: May–September) and a cold season (with data from periods of after the Olympics and the winter period: October–March) according to the previous literature (Ito et al., 2011). HRV indices were log10-transformed to improve the normality and stabilize the variance. Exposure metrics of 5-min, 15-min, 30-min, 1-h, 2-h, 3-h and 4-h moving averages for real-time levels of temperature, relative humidity and air pollutants were used in the present analysis as previously shown to best capture the cardiac effects of environmental exposures on the same study subjects. Log10-transformed 5-min HRV indices were regressed on these moving averages in the mixed-effects models, and typical results based on the exposure metrics that best capture the exposure effects were selected for presentation. Potential confounders including age, time of day, log10-transformed heart rate and relative humidity were included as fixed-effect terms, and day of the year was included as a random-effect term in the models based on our previous analyses (Wu et al., 2010). The following variables were also examined for their potential confounding effects: gender, body mass index, years as a taxi driver, day of week, status of cab windows and status of cab air-conditioner. Regression coefficients of these variables were not statistically significant in any of the models, and therefore these variables were not adjusted in the final models. To account for the nonlinear relationships between weather variables (temperature and relative humidity) and HRV indices, we included both linear and quadratic terms of weather variables in base models (Pope et al., 2004a; Wu et al., 2010). We included a random intercept for each subject and a first-order autoregressive covariance structure in the models to account for the repeated HRV measurements on the same subjects (Delfino et al.,

84

S. Wu et al. / Environmental Research 120 (2013) 82–89

2008). The mixed-effects models were fitted as: Y it ¼ a0 þ bi þ cj þ b1 Temperature þ b2 ðTemperatureÞ2 þ ðb3 PollutantÞþ covariates þ eit

ð1Þ

where Yit is the logarithm of 5-min HRV index for subject i at time t, a0 is the overall intercept, bi is the specific random intercept for the subject i, cj is the random intercept for ‘day of the year’ variable at day j, covariates are potential confounders included as fixed-effect terms (age, time of day, log10-transformed heart rate, and relative humidity in linear and quadratic terms), and eit is the error for subject i at time t. The term for the air pollutant (PM2.5 or CO) was also included in temperaturepollutant joint models. The exposure–response relationships between temperature and HRV indices in different seasons were then fitted using a loess smoother method (Schwartz et al., 2005; Wu et al., 2010). We used 3 degrees of freedom and a smoothing parameter of 0.60 in loess models as previously found to fit the data appropriately (Wu et al., 2010). In order to give a quantitative view over the relationship between temperature and HRV, we performed linear regression analyses to explore temperature effects after stratifying the data according to the temperature cut points which critically determined the shapes of the exposure–response curves. The linear mixed-effects models were fitted as: Y it ¼ a0 þ bi þ cj þ b1 Temperature þ ðb2 PollutantÞþ covariates þ eit ð2Þ To assess the effect modification by traffic-related air pollutants, we further included a term for the interaction between temperature and air pollutant (PM2.5 or CO) in temperature-pollutant joint models. Final results are presented as estimated percent changes in 5-min HRV indices with 95% confidence intervals associated with per interquartile range increase (6.7 1C) in the 5-min average temperature levels over the study.

3. Results 3.1. Subject characteristics, exposure and health data The eligible subjects included 6 males and 8 females, with a mean age of 35.6 years and a mean employment time as a taxi driver of 6.0 years (Table 1). Subjects had a mean body mass index of 25.9 kg/ m2, and were all in good health with blood pressures, resting heart rate, and blood cholesterol, triglyceride and lipoproteins in normal ranges. None of the eligible subjects had a history of smoking or cardiovascular diseases. The overall database of the study included real-time exposure data and HRV data from 48 sampling days, with 26 day and 22 day in the warm and cold seasons, respectively. Among the 14 taxi drivers, ten provided data on 4 day, one on 3 day, two on 2 day and the other one on 1 day. Table 2 summarizes the descriptive statistics on real-time measured environmental exposure and health data by season. Mean 5-min real-time temperature level was higher in the warm season (28.7 1C) than that (24.3 1C) in the cold season, and the temperature range was wide within each season (15.8– 52.4 1C in the warm season and 15.0–40.8 1C in the cold season, respectively). We also obtained data on daily levels of ambient temperature and relative humidity from local ambient weather station (Wu et al., 2011a). The daily mean (standard deviation, range) ambient temperature and relative humidity levels were 23.9 1C (2.3 1C, 17.5 –26.8 1C) and 67.8% (8.0%, 45–82%) in the warm season, and 6.6 1C (4.4 1C, 0.1 –14.9 1C) and 49.3% (15.3%, 21–78%) in the cold season, respectively. Fig. 1 shows the time series of daily levels of in-car temperature and ambient temperature over the study. There were substantial differences between these two temperature measures in the cold season but not in the warm season. 3.2. Exposure–response relationships for temperature and heart rate variability indices Fig. 2 shows the smoothed curves for the exposure–response relationships between personal temperature exposure and 5-min HRV indices by season. We found generally inverse relationships between temperature and standard deviation of normal to normal intervals in both the warm and cold seasons, and the inverse relationships are more obvious in lower temperature ranges

(r27 1C and o25 1C for the warm and cold seasons, respectively). The exposure–response curves for temperature and low-frequency power are similar to the capital ‘V’ in both seasons. In contrast, the curves for temperature and high-frequency power are different in the warm and cold seasons: the curve in the warm season is similar to a skew capital ‘S’ whereas the curve in the cold season is more similar to the curve for temperature and low-frequency power. The exposure–response curve for temperature and low-frequency:highfrequency ratio in the warm season is slightly inverse in lower temperature range (r29 1C) and positive in higher temperature range (429 1C), whereas the curve in the cold season is positive in lower temperature range (r25 1C) and generally even in higher temperature range (o25 1C). These exposure–response relationships were consistent over different exposure metrics and after adjusting for the air pollutants PM2.5 or CO (results not shown).

3.3. Linear temperature effects on heart rate variability indices Table 3 shows the results from linear mixed-effects models after stratifying the data according to the temperature cut points which critically determined the shapes of the exposure–response relationships as shown in the Fig. 2. Generally, we found heterogeneous temperature effects across different temperature strata based on the linear regression analyses. Increasing temperature levels were associated with lower standard deviation of normal to normal intervals across different strata in both seasons, and the temperature effects were stronger in the lower temperature ranges (r 27 1C and r25 1C in the warm and cold seasons, respectively). For example, for an interquartile range increase (6.7 1C) in temperature in the warm season, there were a 22.6% (95% confidence interval: 30.3%, 14.1%) decline and a 2.3% (95% confidence interval: 5.7%, 1.1%) decline in standard deviation of normal to normal intervals in the ‘ r27 1C’ stratum and ‘427 1C’ stratum, respectively. Adjusting for the air pollutants PM2.5 or CO did not change these temperature effects materially. In accordance with the exposure–response curves, we observed opposite linear associations between temperature and lowfrequency power after stratifying the data according to the temperature cut points. There were inverse associations between temperature and low-frequency power in the lower temperature ranges (r25 1C and r26 1C in the warm and cold seasons, respectively) and positive associations in the higher temperature ranges (425 1C and 426 1C in the warm and cold seasons, respectively). Adjusting for CO strengthened the temperature effect in the ‘426 1C’ stratum in the cold season, i.e., for a 6.7 1C increase in temperature, the increase in low-frequency power changed from 13.5% (95% confidence interval:  4.0%, 34.3%) to 22.3% (95% confidence interval: 2.4%, 46.1%) after adjusting for CO in the model. Table 1 Characteristics of the study subjects (n¼14). Characteristics Female (%) Age Mean (years) Range (years) Employed as a taxi driver (years) Body mass index (kg/m2) Seated blood pressure Systolic (mmHg) Diastolic (mmHg) Resting mean heart rate (bpm) Cholesterol (mmol/L) Triglycerides (mmol/L) High-density lipoprotein (mmol/L) Low-density lipoprotein (mmol/L)

8 (57%) 35.67 4.4 27–42 6.07 3.2 25.97 3.4 109.37 10.9 72.97 8.9 63.97 6.6 4.47 0.5 1.27 0.5 1.07 0.2 2.07 0.3

S. Wu et al. / Environmental Research 120 (2013) 82–89

85

Table 2 Descriptive statistics on 5-min averages of in-car real-time exposure variables and 5-min heart rate variability indices by season. Season

N

Mean 7Standard deviation

Temperature, 1C Warm 3065 28.7 74.9 Cold 2540 24.3 73.6 Relative humidity, % Warm 3065 39.9 712.4 Cold 2540 21.5 77.8 PM2.5, mg/m3 Warm 3075 71.3 764.6 Cold 2577 78.9 772.1 CO, ppm Warm 2993 3.8 72.1 Cold 2578 3.0 71.5 Standard deviation of normal to normal intervals, millisecond Warm 2903 47.0 715.6 Cold 2445 44.6 714.4 Low-frequency power, millisecond2 Warm 2868 646.9 7430.3 Cold 2443 574.4 7406.6 2 High-frequency power, millisecond Warm 2863 164.3 7165.3 Cold 2451 151.1 7124.2 Low-frequency:high-frequency ratio Warm 2844 5.7 74.1 Cold 2431 5.0 73.2

Median

25th–75th

Range

28.9 24.1

25.3–31.4 21.8–26.5

15.8–52.4 15.0–40.8

6.1 4.7

38.2 21.1

30.9–47.7 14.8–27.1

12.2–95.3 5.5–51.4

16.8 12.3

53.1 54.8

21.9–103.7 26.1–111.2

1.1–362.1 3.9–566.9

81.8 85.1

2.3–4.9 1.9–3.8

0.2–19.1 0.2–13.9

36–56 34–53

14–140 15–106

544.1 473.7

329.4–852.7 290.4–728.8

32.4–2699.2 41.1–2788.1

523.3 438.4

117.7 116.6

60.9–211.5 55.5–205.5

7.7–1589.0 8.7–810.5

150.6 150.0

2.9–7.3 2.8–6.4

0.4–38.1 0.4–25.1

3.5 2.7 46 43

4.6 4.3

Interquartile range

2.6 1.9 20 19

4.5 3.5

Fig. 1. Time series of daily levels of in-car temperature and ambient temperature over the study. The vertical line separates the data of the warm season and cold season.

Data for high-frequency power in the warm season were divided into three strata according to the ‘S’-like exposure– response curve as shown in the Fig. 2. For a 6.7 1C increase in temperature, we found a strong decline of 17.9% (95% confidence interval: 30.4%,  3.2%) in high-frequency power in the highest temperature range (435 1C) in the warm season. Temperature effects in the other two strata (o25 1C and 25–35 1C) in the warm season were weaker or less consistent after adjusting for CO. Additionally, we found an increase in high-frequency power associated with increasing temperature levels in the higher temperature range (425 1C) in the cold season. This increase in high-frequency power changed from 11.6% (95% confidence interval:  1.0%, 25.8%) to 14.0% (95% confidence interval: 1.0%, 28.6%) per 6.7 1C increase in temperature after adjusting for PM2.5. We found positive linear associations between temperature and low-frequency:high-frequency ratio across different strata except the ‘r25 1C’ stratum in the cold season. In particular, there was a strong increase of 24.5% (95% confidence interval: 15.5%, 34.1%) in lowfrequency:high-frequency ratio per 6.7 1C increase in temperature in the higher temperature range (429 1C) in the warm season.

3.4. Effect modification by traffic-related air pollutants As shown in Table 3, there were significant interactions between temperature and PM2.5 (po0.05 for the interaction terms) on standard deviation of normal to normal intervals across different strata except the ‘r25 1C’ stratum in the cold season (p¼0.22 for the interaction term). We also found a significant interaction between temperature and CO (po0.01 for the interactions term) in the higher temperature range (427 1C) in the warm season. For low-frequency power and high-frequency power, we observed most significant interactions between temperature and PM2.5 (po0.05 for the interaction terms) in the warm season but not in the cold season. We did not find significant interactions between temperature and CO on low-frequency power, but temperature may interact with CO on high-frequency power in the higher temperature ranges (25–35 1C and 435 1C) in the warm season and in the lower temperature range (r25 1C) in the cold season. In addition, temperature may also interact with CO (but not PM2.5) on lowfrequency:high-frequency ratio in the lower temperature ranges (r29 1C and r25 1C in the warm and cold seasons, respectively).

86

S. Wu et al. / Environmental Research 120 (2013) 82–89

Fig. 2. Smoothed curves of the predicted 5-min heart rate variability indices in association with personal temperature exposure based on the mixed-effects models including fixedeffect terms of age, time of day, log10-transformed heart rate, and both linear and quadratic terms of temperature and relative humidity, as well as random-effect terms of subject and day of the year. Series: standard deviation of normal-to-normal intervals in the warm season (A) and cold season (B); low-frequency power in the warm season (C) and cold season (D); high-frequency power in the warm season (E) and cold season (F); low-frequency:high-frequency ratio in the warm season (G) and cold season (H).

3.5. Air pollution effects by temperature level In addition, we re-examined the associations of traffic-related air pollutants with HRV indices after stratifying the data according to the temperature levels (low/high) (Table A2). We found generally stronger associations of these air pollutants with HRV indices in the presence of high temperature levels (Table A3).

4. Discussion 4.1. Summary The cardiovascular effects of ambient temperature exposure have received growing attention in recent studies

(Adamopoulos et al., 2010; Brook et al., 2011; Ren et al., 2006; Stafoggia et al., 2008; Zanobetti and Schwartz, 2008), and ambient temperature and air pollution have been shown to have interactive effect on the occurrence of cardiovascular disease (Ren et al., 2006; Stafoggia et al., 2008). However, few studies have examined the roles of temperature and air pollution simultaneously in the underlying biological mechanisms through which environmental exposure may slead to the cardiovascular disease. In the present study, we reported the relationship of personal temperature exposure with cardiac autonomic function as reflected by HRV in the context of traffic-related air pollution for the first time, and our findings suggest that temperature and air pollution may act together to enhance the cardiac autonomic dysfunction in healthy subjects.

S. Wu et al. / Environmental Research 120 (2013) 82–89

87

Table 3 Estimated percent changes with 95% confidence intervals in 5-min heart rate variability indices per interquartile rangea increase in temperature in linear mixed-effects models by season and temperature category and p-values for the interaction terms in temperature-pollutant joint models. Season

Stratumb

Estimate (95% confidence interval) Base linear model

Standard deviation of normal to normal intervals Warm r27 1C  22.6 (  30.3,  14.1) 427 1C  2.3 (  5.7, 1.1) Cold r25 1C  16.4 (  24.4,  7.5) 425 1C  4.7 (  12.2, 3.6) Low-frequency power Warm r25 1C  6.4 (  26.5, 19.3) 425 1C 12.0 (5.6, 18.8) Cold r26 1C  2.5 (  17.3, 14.9) 426 1C 13.5 (  4.0, 34.3) High-frequency power Warm o25 1C 9.9 (  13.4, 39.4) 25 –35 1C  5.6 (  13.0, 2.4) 435 1C  17.9 (  30.4,  3.2) Cold r25 1C  1.7 (17.1, 16.5) 425 1C 11.6 (  1.0, 25.8) Low-frequency:high-frequency ratio Warm r29 1C 9.0 (  2.3, 21.6) 429 1C 24.5 (15.5, 34.1) Cold r25 1C  10.2 (  23.5, 5.4) 425 1C 9.3 (  4.5, 25.0)

p-value for the interaction term

Adjusting for PM2.5

Adjusting for CO

With PM2.5

 25.7  1.9  16.7  3.8

(  33.6,  16.9) (  5.2, 1.6) (  24.9,  7.6) (  11.5, 4.5)

 13.3  2.4  18.3  5.1

( 22.2,  3.4) ( 5.7, 1.0) ( 26.8,  8.9) ( 12.7, 3.1)

o0.01 o0.01 0.22 0.03

0.59 o 0.01 0.40 0.06

(  23.4, 25.6) (5.5, 18.9) (  18.8, 13.1) (  1.2, 41.2)

 15.3 12.3  10.7 22.3

( 35.3, 11.0) (5.8, 19.1) ( 24.9, 6.3) (2.4, 46.1)

0.04 o0.01 0.65 0.12

0.20 0.63 0.67 0.22

11.1  4.8  15.8 0.2 14.0

(  12.6, 41.2) (  12.5, 3.5) (  28.9,  0.2) (  15.5, 18.8) (1.0, 28.6)

 4.0  5.0  18.4  5.0 9.4

( 33.5, 38.5) ( 12.3, 2.8) ( 30.6,  4.0) ( 20.7, 13.8) ( 3.1, 23.6)

o0.01 o0.01 0.84 0.18 0.87

0.69 o 0.01 o 0.01 0.01 0.12

6.2 25.6  12.2 8.7

(  5.3, 19.1) (16.5, 35.4) (  25.3, 3.1) (  5.2, 24.6)

9.1 25.3  14.6 8.6

( 2.9, 22.6) (16.4, 35.0) ( 27.8, 1.0) ( 5.3, 24.5)

0.62 0.68 0.24 0.72

0.04 0.74 o 0.01 0.53

 1.9 12.0  4.2 18.1

With CO

We used 2-h moving average for standard deviation of normal to normal intervals in both the warm and cold seasons, and 15-min moving average in the warm season and 1-h moving average in the cold season for low-frequency power, high-frequency power and low-frequency: high-frequency ratio. Estimates are adjusted for fixed-effect terms of age, time of day, log10-transformed heart rate, relative humidity, and random-effect terms of subject and day of the year. a b

For comparisons between the warm and cold seasons, we used the interquartile range of 6.7 1C for 5-min average temperature levels over the entire study. Strata were determined according to the temperature cut points as shown in the exposure–response curves in the Fig. 2.

4.2. Exposure measurements in personal levels In the study, we measured the subjects’ real-time, personal exposures to temperature and traffic-related air pollutants to serve as better estimation for their real environmental exposures. Results showed substantial differences between the temperature exposure data in personal levels and data obtained from local weather station, especially in the cold season (Fig. 1). This allowed us to examine the relationship of temperature with cardiac autonomic function as reflected by HRV in a more reliable manner especially in the cold season. In a previous epidemiologic study, daily ambient temperature has been associated with HRV in an aging population in the warm season but not in the cold season (Ren et al., 2011). The authors inferred that ambient temperature would be a poor surrogate for real temperature exposure of the local residents in the winter because most families use heating and residential outdoor activities are very rare during the winter, and therefore estimates based on ambient temperature might be severely biased because ambient temperature and indoor temperature are not correlated in the cold season. In contrast, based on the more accurate temperature exposure data measured in personal levels, we found significant and consistent associations of temperature with HRV in both the warm and cold seasons, providing reliable evidence for the potential role that temperature may play in the cardiac autonomic dysfunction as reflected by changes in HRV. 4.3. Exposure–response relationships for temperature and heart rate variability indices In the present study, we reported the exposure–response relationship for personal temperature exposure and HRV in an epidemiologic scenario for the first time. According to the exposure– response curves, we hypothesized that the study subjects’ HRV may respond to temperature exposure in somewhat different patterns in the warm and cold seasons. The exposure–response curves also

reveal the potential nonlinear relationships between temperature and HRV indices, suggesting a complicated mechanism that human’s autonomic nervous function may respond to the ambient temperature changes. 4.4. Linear temperature effects on heart rate variability indices In the current study, we found consistent inverse associations of temperature with standard deviation of normal to normal intervals across different temperature strata in both the warm and cold seasons (Table 3). These findings suggest general oppressive effect of increasing temperature levels on the overall HRV levels, which is consistent with the previous report (Ren et al., 2011). Temperature effects on HRV indices in the frequency-domain have been examined by several experimental studies (Bruce-Low et al., 2006; Sollers et al., 2002; Yamamoto et al., 2007; Yao et al., 2009). One of these studies examined HRV of 20 healthy subjects at a heated condition of 35 1C for 30 min, and found decreased high-frequency component and increased low-frequency component associated with heat exposure (Sollers et al., 2002). A recent study of six healthy male adults also demonstrated that heat exposure at a high temperature of 35 1C for 30 min could induce activation of the sympathetic nervous system and a withdrawal of the parasympathetic nervous system as evidenced by a significant increase in low-frequency:high-frequency ratio and a significant reduction in high-frequency component (Yamamoto et al., 2007). These findings were duplicated by another experimental study which exposed 10 healthy male subjects to a much higher mean ambient temperature of 74.3 1C (Bruce-Low et al., 2006). We also found positive linear associations of low-frequency power and low-frequency:high-frequency ratio with increasing temperature levels in the higher temperature ranges ( 425 1C or 426 1C or 429 1C) in both the warm and cold seasons (Table 3), which are consistent with the previous experimental studies. These findings suggest an activation of the sympathetic nervous system and a shift towards sympathetic dominance associated

88

S. Wu et al. / Environmental Research 120 (2013) 82–89

with higher temperature exposure. We also found a 17.9% (95% confidence interval:  30.4%,  3.2%) decline in high-frequency power per 6.7 1C increase in temperature in the higher temperature range (435 1C) in the warm season, which may suggest a withdrawal of parasympathetic nervous system under higher temperature levels. It is noteworthy that the temperature levels measured in our study were continuous and wider in ranges which were different from the previous experimental studies using several specific temperature points as the exposure parameters. Nevertheless, the generally consistent findings between our study and the previous experimental studies provide further evidence for the association of temperature and cardiac autonomic dysfunction in human subjects. 4.5. Effect modification by traffic-related air pollutants In the present analysis, we also found significant effect modification of the associations between temperature and HRV indices by two major traffic-related air pollutants (PM2.5 and CO). This is consistent with the previous report by Ren et al. (2011) which found that temperature and ozone may interact synergistically to affect HRV in the study population. Our results further suggest that temperature and air pollution might act together to alter the cardiac autonomic function in human subjects.

taking the advantage of significant ambient temperature changes under natural conditions. First, we provided clues for the responsive patterns of HRV associated with personal temperature exposure by demonstrating detailed exposure–response curves. Second, we found robust linear associations between temperature and HRV indices by stratified analysis. Third, we found significant effect modification of these associations by two major trafficrelated air pollutants (PM2.5 and CO). These findings provide further evidence that autonomic nervous system may be involved in the biological pathways through which environmental exposures (temperature and air pollution) may influence the cardiovascular health synergistically.

Funding sources This work was supported by Grants from the National Key Technologies R&D Program of China [No. 2006BAI19B06], the National Natural Science Foundation of China [No. 81072267], the National High Technology Research and Development Program of China [No. 2012AA062804], and the Academic Award for Excellent Doctoral Candidates of the Ministry of Education of the People’s Republic of China (Shaowei Wu).

Competing financial interests 4.6. Strengths and limitations None declared. The major strength of the study is the repeated-measure design with detailed exposure measurements in personal levels. Subjects were repeatedly studied in different seasons and thus could serve as their own controls well. In particular, the significant changes in personal temperature exposure levels strengthened the study’s ability to assess the exposure–response relationship over a wide temperature exposure range. Furthermore, we recruited nonsmoking, healthy adults free of any cardiovascular compromises as the study subjects to avoid confounding from variations in personal characteristics (e.g., old age, smoking, disease status, medication use, obesity, etc). However, the study also has several limitations. First, the traffic control measures during the study influenced the traffic density and might have introduced a certain confounding effect by influencing the stress level of the drivers (Vrijkotte et al., 2000). We partly controlled this confounding by including a random-effect term for day of the year in the models to account for the autocorrelation between measurements within each day. Second, we only measured real-time levels of two major traffic-related air pollutants (PM2.5 and CO) to represent the traffic-related air pollution, whereas other air pollutants such as ultrafine particles and nitrogen dioxide were not measured. These air pollutants may also play important roles in the cardiovascular effects. For example, quasi-ultrafine particles r0.25 mm in diameter has been shown to have stronger cardiovascular effects than larger particles (Delfino et al., 2008), and both nitrogen dioxide and temperature exposures have been associated with blood pressure changes in pregnant women (Hampel et al., 2011). Third, the temperature effects on HRV as found in this panel of healthy adults may only represent cardiac response in similar population, and should be interpreted with caution when compared with results from other populations with different backgrounds (e.g., susceptible subjects with reduced cardiovascular function). 5. Conclusion In sum, our study provide several novel insights into the relationship between HRV and temperature and air pollution by

Acknowledgments The authors thank C. Yang for conducting the HRV measurements.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2012. 08.008.

References Adamopoulos, D., Vyssoulis, G., Karpanou, E., Kyvelou, S.M., Argacha, J.F., Cokkinos, D., Stefanadisb, C., van de Borne, P., 2010. Environmental determinants of blood pressure, arterial stiffness, and central hemodynamics. J. Hypertens. 28, 903–909. Brook, R.D., Shin, H.H., Bard, R.L., Burnett, R.T., Vette, A., Croghan, C., Williams, R., 2011. Can personal exposures to higher nighttime and early-morning temperatures increase blood pressure? J. Clin. Hypertens. 13, 881–888. Bruce-Low, S.S., Cotterrell, D., Jones, G.E., 2006. Heart rate variability during high ambient heat exposure. Aviat. Space Environ. Med. 77, 915–920. Delfino, R.J., Staimer, N., Tjoa, T., Polidori, A., Arhami, M., Gillen, D.L., Kleinman, M.T., Vaziri, N.D., Longhurst, J., Zaldivar, F., Sioutas, C., 2008. Circulating biomarkers of inflammation, antioxidant activity, and platelet activation are associated with primary combustion aerosols in subjects with coronary artery disease. Environ. Health Perspect. 116, 898–906. Goldberg, M.S., Burnett, R.T., Yale, J.F., Valois, M.F., Brook, J.R., 2006. Associations between ambient air pollution and daily mortality among persons with diabetes and cardiovascular disease. Environ. Res. 100, 255–267. Guo, Y., Barnett, A.G., Pan, X., Yu, W., Tong, S., 2011. The impact of temperature on mortality in Tianjin, China: a case-crossover design with a distributed lag nonlinear model. Environ. Health Perspect. 119, 1719–1725. Hampel, R., Breitner, B., Schneider, A., Zareba, W., Kraus, U., Cyrys, J., Geruschkat, U., Belcredi, P., Muller, M., Wichmann, H., Peters, A., 2012. Acute air pollution effects on heart rate variability are modified by SNPs involved in cardiac rhythm in individuals with diabetes or impaired glucose tolerance. Environ. Res. 112, 177–185. Hampel, R., Lepeule, J., Schneider, A., Bottagisi, S., Charles, M.A., Ducimetie re, P., Peters, A., Slama, R., 2011. Short-term impact of ambient air pollution and air temperature on blood pressure among pregnant women. Epidemiology 22, 671–679. He, F., Shaffer, M.L., Li, X., Rodriguez-colon, S., Wolbrette, D.L., Williams, R., Cascio, W.E., Liao, D., 2011. Individual-level PM2.5 exposure and the time course of

S. Wu et al. / Environmental Research 120 (2013) 82–89

impaired heart rate variability: the APACR Study. J. Expo. Sci. Environ. Epidemiol. 21, 65–73. Ito, K., Mathes, R., Ross, Z., Na´das, A., Thurston, G., Matte, T., 2011. Fine particulate matter constituents associated with cardiovascular hospitalizations and mortality in New York City. Environ. Health Perspect. 119, 467–473. Jia, X., Song, X., Shima, M., Tamura, K., Deng, F., Guo, X., 2011. Acute effect of ambient ozone on heart rate variability in healthy elderly subjects. J. Expo. Sci. Environ. Epidemiol. 21, 541–547. Jia, X., Song, X., Shima, M., Tamura, K., Deng, F., Guo, X., 2012. Effects of fine particulate on heart rate variability in Beijing: a panel study of healthy elderly subjects. Int. Arch. Occup. Environ. Health 85, 97–107. Kan, H., Chen, R., Tong, S., 2012. Ambient air pollution, climate change, and population health in China. Environ. Int. 42, 10–19. La Rovere, M.T., Pinna, G.D., Maestri, R., Mortara, A., Capomolla, S., Febo, O., Ferrari, R., Franchini, M., Gnemmi, M., Opasich, C., Riccardi, P.G., Traversi, E., Cobelli, F., 2003. Short-term heart rate variability strongly predicts sudden cardiac death in chronic heart failure patients. Circulation 107, 565–570. Nolan, J., Batin, P.D., Andrews, R., Lindsay, S.J., Brooksby, P., Mullen, M., Baig, W., Flapan, A.D., Cowley, A., Prescott, R.J., Neilson, J.M.M., Fox, K.A.A., 1998. Prospective study of heart rate variability and mortality in chronic heart failure: results of the United Kingdom heart failure evaluation and prospective study of heart rate variability and mortality in chronic heart assessment of risk trial (UK-Heart). Circulation 98, 1510–1516. Pope III, C.A., Hansen, M.L., Long, R.W., Nielsen, K.R., Eatough, N.L., Wilson, W.E., Eatough, D.J., 2004a. Ambient particulate air pollution, heart rate variability, and blood markers of inflammation in a panel of elderly subjects. Environ. Health Perspect. 112, 339–345. Pope III, C.A., Burnett, R.T., Thurston, G.D., Thun, M.J., Calle, E.E., Krewski, D., Godleski, J.J., 2004b. Cardiovascular mortality and long-term exposure to particulate air pollution: epidemiological evidence of general pathophysiological pathways of disease. Circulation 109, 71–77. Ren, C., O’Neill, M.S., Park, S.K., Sparrow, D., Vokonas, P., Schwartz, J., 2011. Ambient temperature, air pollution, and heart rate variability in an aging population. Am. J. Epidemiol. 173, 1013–1021. Ren, C., Williams, G.M., Tong, S., 2006. Does particulate matter modify the association between temperature and cardiorespiratory diseases? Environ. Health Perspect. 114, 1690–1696. Riediker, M., Cascio, W.E., Griggs, T.R., Herbst, M.C., Bromberg, P.A., Neas, L., Williams, R.W., Devlin, R.B., 2004. Particulate matter exposure in cars is

89

associated with cardiovascular effects in healthy young men. Am. J. Respir. Crit. Care Med. 169, 934–940. Schwartz, J., Litonjua, A., Suh, H., Verrier, M., Zanobetti, A., Syring, M., Nearing, B., Verrier, R., Stone, P., MacCallum, G., Speizer, F.E., Gold, D.R., 2005. Traffic related pollution and heart rate variability in a panel of elderly subjects. Thorax 60, 455–461. Sollers 3rd, J.J., Sanford, T.A., Nabors-Oberg, R., Anderson, C.A., Thayer, J.F., 2002. Examining changes in HRV in response to varying ambient temperature. IEEE. Eng. Med. Biol. 21, 30–34. Stafoggia, M., Schwartz, J., Forastiere, F., Perucci, C.A., Group., SISTI, 2008. Does temperature modify the association between air pollution and mortality? A multicity case-crossover analysis in Italy. Am. J. Epidemiol. 167, 1476–1485. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology., 1996. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Circulation 93, 1043–1065. Vrijkotte, T., van Doornen, L., de Geus, E., 2000. Effects of work stress on ambulatory blood pressure, heart rate, and heart rate variability. Hypertension 35, 880–886. Wu, S., Deng, F., Niu, J., Huang, Q., Liu, Y., Guo, X., 2010. Association of heart rate variability in taxi drivers with marked changes of particulate air pollution in Beijing in 2008. Environ. Health Perspect. 118, 87–91. Wu, S., Deng, F., Niu, J., Huang, Q., Liu, Y., Guo, X., 2011a. Exposures to PM2.5 components and heart rate variability in taxi drivers around the Beijing 2008 Olympic Games. Sci. Total Environ. 409, 2478–2485. Wu, S., Deng, F., Niu, J., Huang, Q., Liu, Y., Guo, X., 2011b. The relationship between traffic-related air pollutants and cardiac autonomic function in a panel of healthy adults: a further analysis with existing data. Inhal. Toxicol. 23, 289–303. Yamamoto, S., Iwamoto, M., Inoue, M., Harada, N., 2007. Evaluation of the effect of heat exposure on the autonomic nervous system by heart rate variability and urinary catecholamines. J. Occup. Health 49, 199–204. Yao, Y., Lian, Z., Liu, W., Jiang, C., Liu, Y., Lu, H., 2009. Heart rate variation and electroencephalograph—the potential physiological factors for thermal comfort study. Indoor Air 19, 93–101. Zanobetti, A., Gold, D.R., Stone, P.H., Suh, H.H., Schwartz, J., Coull, B.A., Speizer, F.E., 2010. Reduction in heart rate variability with traffic and air pollution in patients with coronary artery disease. Environ. Health Perspect. 118, 324–330. Zanobetti, A., Schwartz, J., 2008. Temperature and mortality in nine US cities. Epidemiology 19, 563–570.