Exposure of chronic obstructive pulmonary disease patients to particles: Respiratory and cardiovascular health effects

Article metrics

Abstract

To examine hypotheses regarding air pollution health effects, we conducted an exploratory study to evaluate relationships between personal and ambient concentrations of particles with measures of cardiopulmonary health in a sample of patients with chronic obstructive pulmonary disease (COPD). Sixteen currently non-smoking COPD patients (mean age=74) residing in Vancouver were equipped with a particle (PM2.5) monitor for seven 24-h periods. Subjects underwent ambulatory heart monitoring, had their lung function and blood pressure (BP) measured, and recorded symptoms and medication use. Ambient PM2.5, PM10, sulfate, and gaseous pollutant concentrations were monitored at five sites within the study area. Although no associations between air pollution and lung function were statistically significant, an estimated effect of 3% and 1% declines in daily FEV1 change (ΔFEV1) for each 10 μg/m3 increase in ambient PM10 and PM2.5, respectively, was observed. Increases of 1 μg/m3 in personal or ambient sulfate were associated with 1.0% and 0.3% declines in ΔFEV1, respectively. Weak associations were observed between particle concentrations and increased supraventricular ectopic heartbeats and with decreased systolic BP. No consistent associations were observed between any particle metric and diastolic BP, heart rate, or heart rate variability (r-MSSD or SDNN), symptom severity, or bronchodilator use. Of the pollutants measured, ambient PM10 was most consistently associated with health parameters; the use of personal exposures did not improve the strength of any associations or lead to increased effect estimates.

Introduction

Many epidemiological studies have indicated that cardiopulmonary morbidity and mortality are increased during, or just after, periods of elevated particulate matter (PM) levels. Time series studies (Pope et al., 1992; Schwartz, 1994a,b, 1997, 1999; Schwartz and Dockery, 1992; Borja-Aburto et al., 1998) of acute impacts have been complemented by evidence from prospective cohort studies (Dockery et al., 1993; Pope et al., 1995a,b; Abbey et al., 1998). To support these epidemiological associations, much current research is focused on the elucidation of biologically plausible mechanisms through which airborne particles may act (NRC, 1998).

Here we study the respiratory and cardiovascular status of individuals with chronic obstructive pulmonary disease (COPD) in relation to particle exposure. Associations between morbidity or mortality and ambient PM appear to be due, in part, to elevated risk among the elderly and for those with cardiovascular or respiratory disease, such as COPD (Schwartz, 1994ab; Schwartz et al., 1996). Airway deposition of particles is enhanced in patients with COPD (Bennett et al., 1996). The combination of greater effective dose in the portion of the lung that remains functional and decreased functional reserve capacity may exacerbate these patients' respiratory response to particles (Bates, 1992). Additionally, hypoxemia in COPD could have adverse effects on other organs, including the heart. The mechanism of adverse particle effects in individuals affected by COPD could involve a cardiac pathway that is equally or more important than mechanisms that directly involve the lungs.

In this exploratory study, a variety of non-invasive indicators were used to assess the health status of the study population, including spirometry to assess pulmonary function, and blood pressure (BP) and electrocardiograms (ECGs) to assess cardiovascular function. ECG waveform changes have previously been demonstrated in dogs (Godleski et al., 2000) as well as in rats (Watkinson et al., 1998; Watkinson et al., 1999; Gordon et al., 2000) following particle exposure. An increased occurrence of cardiac arrhythmic events with exposure to particulate air pollution may demonstrate stress on the myocardium due to hypoxemia or increased blood viscosity. Such stress may also be reflected in subjects' heart rate (HR) and heart rate variability (HRV). High HR has been associated in recent studies with hypertension (Gillum, 1988), coronary heart disease (Dyer et al., 1980), and mortality (Goldberg et al., 1996). HRV describes the beat-by-beat changes in lengths of the intervals between beats, and has been used to assess the autonomic effects of drugs, exercise, and other stresses. Decreased HRV marks decreased parasympathetic and increased sympathetic tone (Kleiger et al., 1995). HRV has been shown to be decreased in patients with heart failure (Casolo, 1995) and in postmyocardial infarction patients (Bosner and Kleiger, 1995). Decreased HRV after acute myocardial infarction is also a risk factor for subsequent morbidity and mortality (Singer and Ori, 1995; Vybiral and Glaeser, 1995). Observations of altered HR or HRV in relation to particulate air pollution would help piece together the causal chain linking air pollution, alterations in blood parameters, and heart function.

In this study, we conducted repeated measurements of ambient air pollution and personal particle exposure in combination with spirometry, ECGs, and BP in a population of COPD patients. Detailed individual exposure and health assessments for a susceptible group were unique aspects of this study. Our main objective in this exploratory study was to assess the relative ability of different particle exposure metrics, presumably exhibiting different degrees of exposure misclassification, as indicators of respiratory and cardiovascular health outcomes.

We have previously reported on the relationship between ambient concentrations and personal exposures over time for this study population (Ebelt et al., 2000). Regarding exposure, we found a moderate correlation between personal exposure and ambient PM2.5 (median r of individual regressions=0.48). The correlation with ambient PM2.5 was high, however, when using sulfate (SO42−) as the exposure metric (median r=0.96). This high correlation suggests that personal SO42− is a good marker of exposure to the ambient component of PM. In contrast, the low correlation between personal and ambient PM2.5 results from PM2.5 exposure being comprised of exposure to indoor sources of PM2.5 and PM2.5 from outdoor sources. As epidemiological studies based on ambient particle measures have been associated with respiratory and cardiovascular health outcomes, it was our prior hypothesis that personal SO42− exposures would show the most consistent relationship with heath outcomes as this metric incorporated the least amount of exposure misclassification.

Measurement methods

The study was conducted from April 21 to September 25, 1998. Ethical approval was obtained from the Clinical Research Ethics Board, University of British Columbia. The target population was 25–30 participants based on the results of previous studies and logistical constraints. Study inclusion/exclusion criteria specified individuals with physician-diagnosed light-to-moderate COPD (as defined by FEV1≥0.75 l), aged 60 or above, living within the Greater Vancouver Regional District (GVRD) (excluding the North Shore), and currently not smoking nor living with current smokers.

Exposure measurements have been described in detail elsewhere (Ebelt et al., 2000). Briefly, ambient PM10 (TEOM) and PM2.5 (Harvard Impactor, 4 l/min) concentrations were measured at five sites within the GVRD. Daily 24-h average ambient temperature (T), relative humidity (RH), carbon monoxide (CO), and daily maximum ozone (O3) data were obtained from the GVRD Air Quality Department. Average concentrations from five monitoring sites within the study region were used. In addition, all subjects underwent seven 24-h sampling sessions throughout which personal PM2.5 exposure measurements were collected (Personal Exposure Monitor, 4 l/min). The sampling sessions for each subject were randomly spaced with at least 1.5 weeks between consecutive sessions, and sampling days were distributed evenly within the study period. All filters from ambient and personal monitoring were analyzed for particle mass as well as particulate sulfate. Each subject underwent the seven planned sampling sessions, yielding a 100% compliance rate. Six (5%) of 112 personal exposure samples were excluded due to high or low flow rates or a sample duration less than 20 h. For the 106 valid samples, the number of valid exposure measurements for each subject ranged between five and seven.

Pulmonary function was measured at the beginning and end of each sampling session using portable pneumotach spirometers (Presto FLASH portable spirometer; Spacelabs Medical, Deerfield, Wisconsin). Spirometers were calibrated once every morning on each sampling day using a 3-l syringe. Each subject was asked to perform a minimum of three acceptable forced vital capacity maneuvers (ATS, 1995), while seated and wearing nose plugs. Maneuvers with unsatisfactory starts, coughing during the first second, Valsalva maneuvers (glottis closure), early terminations of expiration, leaks, and obstructed mouthpieces were considered unacceptable. Pulmonary function testing was conducted at the same time (±1 h) on each measurement day.

BP measurements, using a BP cuff/stethoscope combination (Sprague/Rapapport), were conducted at the start and end of each sampling period. Measurements were taken from each subject's left arm, while seated with their elbow resting at mid chest level. For some subjects on some occasions, the diastolic BP could not be identified. In these cases, the diastolic pressure was taken to be 0 mm Hg.

Heart rhythm was measured throughout the 24-h sampling period using Holter monitors (DM-400 cassette Holter recorder). These were battery-operated ambulatory ECG recording devices with three independent ECG channels. At night, subjects placed the Holter recorders beside them in their beds. Subjects were asked not to bathe or shower during the sampling periods.

At the end of each sampling period, subjects completed an eight-question symptom questionnaire. Seven questions addressed symptoms of coughing, sputum production, breathing difficulty, chest pain, heartbeat irregularities, fatigue, and dizziness. Possible responses were whether the respective symptom over the previous 24 h was “more than,” “less than,” or “about the same as” the normal level for that subject. An eighth question recorded the amount of bronchodilator use (i.e., salbutamol and/or ipratropium bromide) by counting the number of times the medication was used during the sampling session. A list of additional medications was compiled for each subject during their first sampling session, indicating type, dose, and frequency of use. At the conclusion of every subsequent sampling period, the medication form was discussed with the subject to determine which medications had been taken during the 24-h sampling period and whether their medication list was still accurate.

Data analysis

Spirometry, BP, and ECGs

When selecting the appropriate forced expiratory maneuvers for the data analysis, the American Thoracic Society (ATS) criteria (1994 update) were used (ATS, 1995). Ninety-nine percent (223 subject sessions) of all FEV1 values used were reproducible according to ATS spirometry criteria (ATS, 1995). Ninety-seven percent (218 subject sessions) of all FVC (forced vital capacity) values were considered reproducible. Four to seven valid BP measurements were obtained per subject, with 98 samples in total. This BP sample number was lower than the number of personal samples (N=106) because BP measurements were not conducted during the first few weeks of the study.

Arrhythmia, HR, and HRV data were obtained by analysis of the Holter recordings on a Biomedical Systems Century Advanced Holter System using superimposition and page mode. A certified Holter scanning technologist sorted the beats into various templates (classifications) (i.e., normal, supraventricular, ventricular, paced, artifact, etc.). Time domain analysis was used to determine HRV. This method is based on different ways of measuring the standard deviation (SD) of sinus (i.e., non-arrhythmic) R–R intervals (measured in milliseconds) over the sampling period. Of the 104 sample sessions in which Holter monitoring was conducted, 102 records of sufficient length (i.e., greater than 20 h) were obtained. Holter monitoring was not conducted during the first eight sampling sessions of the study. Of these 102 measurements, 86 were used in the data analysis, while 16 recordings with excessive background noise (due to tape stretch from previous use) were excluded. In total, three to seven valid ECG recordings were obtained per subject.

The number of potential variables in the analysis was reduced by eliminating one of any two highly correlated variables and by choosing specific variables to represent collections of similar outcomes. High correlations between the pre- and postsampling results for FEV1 (r=0.95), FVC (r=0.94), and FEV1/FVC (r=0.98) were found. We restricted analyses to two measures of FEV1 as it is less effort-dependent than FVC: postsample FEV1 and the change in FEV1 (ΔFEV1=postsample FEV1−presample FEV1). Correlations among the four BP variables indicated that pre- and postsample results were also highly correlated for both systolic (r=0.84) and diastolic BP (r=0.92). Postsample BP measurements were chosen for further analysis as we were primarily interested in evaluating the impact of air pollution on the BP level, rather than on the within-session change in BP. Some potential ECG variables were not analyzed due to lack of variability between subjects. From the remaining variables, 24-h average HR was chosen as the basic rhythm variable. SDNN (standard deviation of normal–normal beats), derived from direct measurements of the beat-to-beat intervals, was chosen as the variable for assessing overall HRV. r-MSSD (root mean square successive difference), derived from the differences between adjacent intervals, was chosen as a measure of short-term HRV. Finally, SVE (supraventricular ectopy) was selected as the arrhythmia variable.

Most variables were not normally distributed, nor was normality improved with log transformation. The distribution of SVE was right-skewed and log-transformed values were used in the analysis. Variables were summarized both by subject and by pooling over all subjects. To control for systematic differences in the levels of responses across individuals, each subject's result for a sampling session (for all variables except ΔFEV1) was standardized by calculating the deviation from that subject's average value across all sampling sessions. Cases with large deviations (>2–4 SD) from subjects' mean values were excluded. One data point was excluded for each of lung function, diastolic BP, and SDNN data sets. Assessment of trends over time showed one subject's postsampling FEV1 to increase considerably throughout the study. This subject also had atrial fibrillation. Therefore, this subject was excluded from all post-FEV1, HR, SDNN, and r-MSSD analyses.

After data clean-up, the eight health endpoints (FEV1, ΔFEV1, systolic and diastolic BP, SVE, HR, SDNN, r-MSSD) were used as dependent variables in pooled ordinary least squares (OLS) regressions against the exposure metrics (ambient PM10, ambient PM2.5, ambient SO42−, personal PM2.5, and personal SO42−). The correlation structures of the endpoints were assessed to determine the presence of autocorrelation between sampling sessions. Since no consistent pattern of autocorrelation was found, no adjustment for such autocorrelation was made in the analyses (recall that there was a 1.5-week period between sampling sessions). Regression diagnostics often indicated unequal variances, which were accounted for in final models by running pooled weighted least squares (WLS) regressions (by the inverse of the estimated variances for individual subjects, as estimated from the residuals of the OLS fit).

The WLS regressions were further analyzed for potential confounding by several meteorological and co-pollutant variables as well as by bronchodilator use (described below). After assessing the correlation between particulate exposure metrics and the potential confounders, two-variable models were used to assess the effect of these secondary variables on the relationships between particles and health endpoints. The two-variable models were fit using weights estimated from the residuals of the OLS regressions between the health indicator and the exposure metric of each specific two-variable model.

Symptoms and Bronchodilator Use

All 16 subjects answered a symptom questionnaire (asking subjects to describe their symptoms on that day as more, less, or the same as most days) at the end of each of the seven sampling sessions, resulting in 112 completed questionnaires. Subjects' answers were organized into two categories: “more” and “same or less.” Questions concerning chest pain, heartbeat irregularities, and dizziness were eliminated from the analyses due to lack of variability in responses. Another category, “any respiratory symptom,” was created by collapsing the responses from the coughing, sputum, and breathing difficulty variables. Responses for this category were considered to be “more” if there was an increase in any of the three symptoms. Both pooled and individual logistic regressions were conducted to evaluate the relationships between the symptoms and exposure measures.

All 16 subjects provided data at the end of each sampling session as to the number of times a bronchodilator had been used during the previous 24 h. To study the relationship between personal exposure to PM2.5 and total bronchodilator use and extra (usage above prescribed dosage) bronchodilator doses, individual regression analyses were conducted and scatterplots were examined.

Results

Recruitment yielded 17 subjects, one of whom was eventually excluded from the data analysis as she did not meet the inclusion criteria for disease status. Thus, the study population consisted of 16 subjects (seven males, nine females) with moderate (forced expiratory volume in 1 s (FEV1) ≥0.75 l) physician-diagnosed COPD. For the potential subjects who were contacted, the following reasons were given for not participating in the study: lack of interest, too much effort required, too busy, or inability to meet the study inclusion criteria. All subjects (mean age 74; range 54–86 years) were residents of the GVRD and all were current non-smokers and did not live with smokers, with the exception of one subject. For this subject, the smoker agreed to smoke outdoors or not at all on each of the subject's sampling days.

Particle concentrations and interquartile ranges (IQR) were relatively low (Table 1). Personal SO42−, ambient SO42−, and ambient PM2.5 were all highly correlated with each other, whereas personal PM2.5 was poorly correlated with the other exposure variables (Table 2). Table 3 summarizes the health measurements, and effect estimates (based on single-pollutant models) for IQR increases in exposure are presented in Table 4.

Table 1 Ambient concentrations and personal exposures; ambient samples are the average of all five sites over each sampling day.
Table 2 Pearson correlation coefficients among particle exposure metrics, co-pollutants, meteorological variables (T, RH), and bronchodilator use (BD use).
Table 3 Summary statistics of respiratory and cardiovascular health outcome measurements; bph=beats per hour; bpm=beats per minute; ms=milliseconds.
Table 4 Effect estimates and standard errors (SE, in parentheses) for IQR increases in pollutant concentration exposure.

Although not statistically significant, the size of the ambient PM10 effect estimate for ΔFEV1 (per IQR increase) was larger than the effect estimates for the other particle measures (Figure 1). An inverse association with a smaller effect estimate (per IQR) was found between personal sulfate exposures and ΔFEV1 over the 24-h sampling periods (−0.012093 ml/μg/m3, P=0.08) (Figure 1), although this relationship was sensitive to the inclusion of O3 in the model (personal SO42− −0.004346 ml/μg/m3; O3 −0.889024 ml/ppb). Estimates of effect from weighted regression analyses indicated that an IQR (5.5 μg/m3) increase in ambient PM2.5 was associated with a 6.8-ml decrease in ΔFEV1. For comparison to other studies, this estimated effect magnitude corresponds to a 1.1% decline in ΔFEV1, relative to the average FEV1, for each 10 μg/m3 increase in ambient PM2.5. IQR (0.9 and 1.2 μg/m3 for personal and ambient sulfate, respectively) increases in personal and ambient sulfate were associated with 10.9 and 3.7 ml decreases in ΔFEV1, respectively. These estimated effects correspond to 1.0% and 0.3% declines in ΔFEV1 for each 1 μg/m3 increase in personal or ambient sulfate, respectively. None of these effect estimates was statistically significant.

Figure 1
figure1

Lung function (ΔFEV1) response to IQR increases in exposure.

No consistent associations were observed between any exposure measure and symptom severity (results not shown). Regarding bronchodilator use, nine subjects used their medication according to a schedule and as needed to alleviate discomfort. Two subjects used the medication only as needed, while two other subjects used the medication only according to a schedule. For three subjects, the bronchodilator prescriptions were altered within the study period. From individual regression analyses, a lack of within-subject variability was observed, which precluded further analyses of this variable.

For cardiovascular endpoints, single-pollutant models (Table 4) indicated that both systolic (Figure 2) and diastolic BP decreased with increasing exposures, although none of the effect estimates for diastolic BP was statistically significant. In two-variable models predicting systolic BP, CO reduced the size of the effect estimates for all exposure indicators with the exception of personal PM2.5; the sign of the effect estimates remained unchanged with the addition of CO. For example, the effect estimate of −0.576 mm Hg/μg/m3 for personal SO42− was reduced to −0.304 mm Hg/μg/m3 once CO was included in the model. After inclusion of other covariates (T, RH, O3, and BD), BP effect estimates remained mostly negative. Models for other cardiovascular outcomes were not sensitive to the inclusion of CO. Increased SVE arrhythmic beats were associated with increases in all exposure metrics (Figure 3), and the magnitude of the effect estimates was not greatly altered by inclusion of covariates in two-pollutant models. HR, SDNN, and r-MSSD regressions resulted in particle effect estimates that were mostly not significant and inconsistent in size and sign. Generally, trends were for HR to increase and HRV to decrease with increasing PM exposures, whereas HR decreased and HRV variables increased with increasing sulfate exposures (Figures 4 and 5). These relationships were largely inconsistent and unstable following addition of covariates.

Figure 2
figure2

Systolic BP response to IQR increases in exposure.

Figure 3
figure3

SVE response to IQR increases in exposure.

Figure 4
figure4

HR response to IQR increases in exposure.

Figure 5
figure5

r-MSSD response to IQR increases in exposure.

Overall, for both lung function and cardiovascular endpoints, ambient PM10 consistently had the largest effect estimates per IQR on the health indicators analyzed. Models using personal exposure measurements did not show larger or more consistently positive effect estimates relative to those using ambient exposure metrics.

Discussion

This study of air pollution effects in patients with COPD is the first to examine multiple respiratory and cardiovascular endpoints in relation to an array of both personal and ambient particle concentrations. Our aim was to determine whether we detect either respiratory or cardiovascular effects in this susceptible subgroup of the population, and whether various measures of particle exposure, presumably exhibiting different degrees of exposure misclassification, differed in their ability to detect effects.

Respiratory Health Endpoints

Though not statistically significant, estimated declines in ΔFEV1 of 3% and 1% for each 10 μg/m3 increase in PM10 and PM2.5, respectively, were observed in subjects with COPD. Similarly, 1 μg/m3 increases in ambient and personal sulfate were associated with declines in ΔFEV1 of 0.3% and 1.0%, respectively. The published literature indicates that physiologically small but statistically significant declines in lung function are associated with ambient particle concentrations (Pope et al., 1995a,b). Previous studies have shown decreases of up to 0.4% in FEV1 associated with each 10 μg/m3 increase in daily mean PM10 (Dockery and Pope, 1996). The comparatively (on a μg/m3 basis) large FEV1 effect estimates observed our study may not be surprising considering that the observed results were based on a susceptible population.

Calculations revealed that for our sample size and variance, changes in FEV1 would have needed to be at approximately 1.6 times as large to have reached statistical significance (for PM10). This is likely due to the low level of pollutant variability in combination with the small number of measurements. Furthermore, studies that have found significant declines in lung function in association with particles were conducted at mean ambient concentrations much greater than those observed in this study, which could have contributed to the lack of significant effects. The mean ambient PM10 concentration in our study was 18 μg/m3. Pope and Kanner (1993) analyzed PM10 levels at 100 μg/m3 and found 2% declines in FEV1 of mild to moderate COPD patients. At 24-h PM10 concentrations exceeding 150 μg/m3, up to a 7% reduction in lung function has been observed (Pope et al., 1995a,b). In other studies, mean PM10 concentrations have ranged from 27 to 76 μg/m3 (Vedal, 1997). Our results were also sensitive to the inclusion of ozone in the regression models. Limitations of sample size and high correlations among the pollutants (r0.6), however, made it difficult to attempt to identify which pollutant(s) was independently associated with the lung function measures.

No consistent results were found between respiratory symptoms and exposure. Previous studies have often reported statistically significant associations between particulate air pollution and respiratory symptoms. Typically, these studies have found that a 10 μg/m3 increase in PM10 has been associated with a 1–15% increase in symptoms (Pope et al., 1995a,b; Dockery and Pope, 1996). In our study, analyses of symptoms were difficult to conduct because of the lack of variability in responses in some individuals.

Cardiovascular Health Endpoints

The mean BP (systolic/diastolic in mm Hg) in our study population was 134/67. These values are similar to those of subjects of similar ages (age 65–89), with and without cardiovascular disease (132/78) (Liao et al., 1999) but higher than those of healthy 22- to 34-year-old adults (110/59) (Hausberg et al., 1997). In relation to particles, we found decreases in systolic BP to be associated with increased exposure. This relationship was consistent across all exposure metrics analyzed and weakly statistically significant (P<0.10) for ambient PM10 and personal PM2.5. Effect estimates describing the relationship between air pollution and systolic BP were, except for personal PM2.5, altered in magnitude by the inclusion CO in two-pollutant models. In these models, CO was negatively associated with BP, which appears to be consistent with other studies of CO (Penney and Howley, 1991). However, the sign of the particle effect estimates in the systolic BP models particles was mostly unchanged with the addition of CO. Our results are in contrast with limited studies that have demonstrated increases in BP upon exposure to PM2.5 in elderly individuals (Gold et al., 2000) or to PM10 in COPD patients (Linn et al., 1998). Increases in BP in response to increases in particles might occur if their deposition resulted in hypoxemia or increases in inflammatory mediators that might lead to increases in blood viscosity.

The arithmetic mean number of SVE events was 36.9 beats/h in this study population. Both supraventricular and ventricular cardiac arrhythmias are common in COPD patients (Gorecka, 1997). Physiologic abnormalities associated with COPD and medications used in the management of the disease are thought to be causes of the arrhythmias in these patients. We found an increase in SVEs associated with both particles and sulfate that was not affected by inclusion of additional variables in models. In the study of Linn et al. (1998), SVEs in COPD patients were also found to increase with outdoor PM10 concentrations. For each 7 μg/m3 IQR increase in ambient PM10 exposure, we estimated a 10.5% (exp[0.1]) increase in SVEs, corresponding to a 3.5 beats/h increase from the population mean number of SVE beats per hour (Table 3). The clinical significance, if any, of this magnitude of effect is not clear. In two studies, each following 64–65 COPD patients, a high prevalence of cardiac arrhythmias was found (including SVEs), although arrhythmias were not found to be significant predictors of death (Shih et al., 1988; Lewczuk et al., 1992). However, observing rats postexposure, Watkinson et al. (1998) found some rats dying abruptly due to “the initiation and propagation of an ectopically stimulated depolarization which `captured' the heartbeat, producing a fatal arrhythmia” upon exposure. If one arrhythmic beat can cause mortality, any increase in arrhythmias due to particle exposure may prove to be serious.

The mean HR for our study population was 81 beats/min. This value is higher compared to those of healthy 22- to 24-year-old adults (HR: 57–61 beats/min) (Hausberg et al., 1997) and elderly subjects ages 65–89 (HR: 65–81 beats/min) (Liao et al., 1999). Our analyses indicated nonsignificant positive effect estimates relating ambient PM and HR, which compare well with other studies reporting positive relationships (Peters et al., 1999; Pope et al., 1999a,b). The sizes of the effect estimates for our data, however, were larger, indicating that a 100 μg/m3 increase in PM10 would lead to a 6 beats/min increase in HR. Pope et al. (1999a,b) estimated that a 100 μg/m3 increase in previous day PM10 was associated with an average increase of 0.78 beats/min. An increase of 200 μg/m3 SO2 was associated with increases of 2.3–2.8 beats/min in the Peters et al. (1999) study. Increased HR may be a significant indicator of mortality in COPD patients with severe hypoxemia (Shih et al., 1988) and thus may also be considered an adverse effect of exposure to ambient PM. The other exposure metrics in our study had negative relationships with HR. Decreased HR has been observed in some rat studies following exposure to ROFA particles (Campen et al., 1996; Watkinson et al., 1998). For example, rats have demonstrated evidence of progressive hypoxic failure of the myocardium with decreased HR and various ECG waveform abnormalities (Watkinson et al., 1998). Thus, if decreased HR is a characteristic of myocardial hypoxia, perhaps such a response to particle exposure for those with preexisting disease might be expected.

For HRV, our subjects had a mean SDNN of 97 ms and a mean r-MSSD of 44 ms. Compared to healthy populations, our subjects had lower mean SDNN and similar or slightly higher mean r-MSSD values (Task Force, 1996; Jensen-Urstad et al., 1997). Decreased R–R variability in COPD patients has been associated with decreased FEV1. The relationships with the HRV responses were inconsistent among the various exposure metrics tested. Most SDNN and r-MSSD relationships decreased with increasing PM exposure and increased with increasing sulfate exposure. Temperature and CO appeared to modify slightly the r-MSSD relationships with PM; however, the sign of the relationships remained negative.

Studies assessing time domain HRV indices in relation to particle exposure have reported mixed results. SDNN has generally been found to decrease in relation to PM exposure (Liao et al., 1999; Pope et al., 1999ab; Gold et al., 2000). Gold et al. (2000) found an effect estimate for SDNN of −0.833 for 17 μg/m3 IQR increase in PM2.5. This compares well with the results of the present study, where a 17.3 μg/m3 increase in exposure, which was the difference between the maximum ambient level and the study mean, yielded an effect estimate of −0.9. The results for r-MSSD have been inconsistent in the few studies analyzing this outcome, with one study reporting a negative relationship (Gold et al., 2000) and another study reporting a positive relationship (Pope et al., 1999a,b).

Comparing Effect Estimates for Different Exposure Metrics

A key research question in studying the health effects of particulate air pollution involves a determination of the component or components of PM that are most closely related to the observed acute PM–mortality associations (Thurston, 1996). In this study, the contribution of indoor source particles and the lack of correlation between personal and ambient measures of PM2.5 (Ebelt et al., 2000) suggested that personal PM2.5 exposure data would not be expected to be highly related to health outcomes. Indeed, the personal PM2.5 effect estimates were generally smaller than all other exposure metrics for all health indicators analyzed. An unexpected finding was that for most relationships, the largest effect estimates were found for models with ambient PM10 as the exposure metric. It is possible that this finding may be an artifact of inadequate control of meteorological factors; while we did control for both temperature and relative humidity, other uncontrolled factors may have affected the results. Also, the ability to adequately control for effects of meteorology with our small sample size may have been limited.

Limitations

Though unexpected results were found, this study may serve as guidance for future studies of similar type. We were disappointed in not being able to recruit a larger number of COPD patients. The major factors, which led to this small sample size, were a lower-than-expected initial pool of applicants who attended clinics and support groups, potential subjects who did not meet the study inclusion criteria, and lack of willingness to participate in the personal monitoring component of the study. Based on our findings and effect estimates for a sample size of 16, a sample size of approximately 40 subjects would have been necessary for effects to be statistically significant.

The very low levels of ambient particles and the low variability in these levels may have also reduced our ability to observe associations between air pollution and health outcomes. Since this study included personal exposure measurements and since we aimed to increase the variability in measured personal exposures, consecutive daily samples were not collected. Further, it is difficult to request subjects to conduct personal monitoring for seven or more consecutive days as the monitoring is often disruptive. Because of this constraint, we were unable to conduct any analyses that evaluated lagged exposures. Further, we could not assess the impact of the previous day's exposure on the baseline lung function measurements, as we have done in previous studies (Brauer et al., 1996).

An important limitation in predicting cardiopulmonary health effects in COPD patients is that, as is often the case, many other factors affect the cardiovascular measured endpoints. Compared with these risk factors, pollution will likely have only a small impact. As an example, our ability to control for medication use was poor. There were many medications other than bronchodilators that were used by our study population, which may also have affected cardiovascular parameters. Future studies may wish to contemplate, prior to data collection, how medication information will be used in the later analysis and in what form this information could be collected to increase its accuracy and usefulness in this analysis.

We experienced a similar problem with the acute respiratory symptom data. We asked subjects questions regarding symptoms on the day of measurement in relation to their “usual level of symptoms as a means to standardize for between-individual variability.” However, by asking whether symptoms were better, worse, or the same as other days, we may have limited the variability in this response. Future studies may consider acute symptoms questions, which can be answered on an ordinal scale with multiple categories of severity. An additional limitation in our study was that day-to-day changes in HRV were not well characterized, and in this type of study where subjects were moderately active, HRV analysis may be too sensitive a measure to use. It may be more feasible to use resting HRV data if these types of data are needed.

Conclusions and implications

This study was exploratory both with respect to determining the health effects of particles and with respect to study design issues. Results suggest that exposure to all particle metrics evaluated, with the exception of personal PM2.5, may be associated with decreased FEV1 in patients with mild to moderate COPD. Symptom analyses did not provide substantial additional information regarding the relationship between exposure to particulate air pollution and respiratory health. In this study population, BP decreased and SVEs increased with increasing particle exposures. The stability of these results suggests BP and SVE to be sensitive indicators of response in COPD patients. HR and HRV were less stable and may not be as sensitive to indicate changes in cardiovascular parameters. The use of “sensitive indicators” may be the preferred objective in future studies. However, not only the direction of the effect, but also the biomedical impact of these indicators on COPD patients remains to be investigated.

In terms of study design, a more sophisticated approach was taken compared to large epidemiological studies. Instead of assessing exposure of the study population using ambient measurements, personal exposure monitoring was conducted. We also obtained detailed health assessments and individual characteristics data. Further, we studied a group of individuals thought to be susceptible to the health effects of particles. Even though our study population was much smaller than most epidemiological studies and the number of repeat measurements was low, the above improvements in study design were made with the aim of clarifying questions regarding the respiratory and cardiovascular health effects of exposure. However, some unexpected results occurred with respect to the directions of the exposure–cardiovascular response relationships and in comparisons of effect estimates from different exposure metrics. The complexity of studying cardiovascular health effects and their potential relationship to air pollution exposure, and the lack of consistency observed in the results of this study suggest that the improvements in our study design have not clarified these questions. More insight is needed on how the complexities of this research can be dealt with.

References

  1. Abbey, DE, Burchette, RJ, Knutsen, SF, McDonnell, WF, Lebowitz, MD, and Enright, PL . Long-term particulate and other air pollutants and lung function in nonsmokers. Am J Respir Crit Care Med (1998) 158: 289–298

  2. ATS. Standardization of spirometry, 1994 update. American Thoracic Society. Am J Respir Crit Care Med (1995) 152: 1107–1136

  3. Bates, DV . Health indices of the adverse effects of air pollution: the question of coherence. Environ Res (1992) 59: 336–349

  4. Bennett, WD, Zeman, KL, and Kim, C . Variability of fine particle deposition in healthy adults: effect of age and gender. Am J Respir Crit Care Med (1996) 153: 1641–1647

  5. Borja-Aburto, VH, Castillejos, M, Gold, DR, Bierzwinski, S, and Loomis, D . Mortality and ambient fine particles in southwest Mexico City, 1993–1995. Environ Health Perspect (1998) 106: 849–855

  6. Bosner, MS, Kleiger, RE . Heart rate variability and risk stratifaction after myocardial infarction. In: Malik M, Camm AJ, (Eds.) Heart Rate Variability. Futura Publishing Company, Inc., Armonk, New York 1995 pp. 331–340

  7. Brauer, M, Blair, J, and Vedal, S . Effect of ambient ozone exposure on lung function in farm workers. Am J Respir Crit Care Med (1996) 154: 981–987

  8. Casolo, GC . Heart rate variability in patients with heart failure. In: Malik M, Camm AJ, (Eds.) Heart Rate Variability. Futura Publishing Company, Inc., Armonk, New York 1995 pp. 449–465

  9. Campen, MJ, Watkinson, WP, Lehmann, JR, and Costa, DL . Modulation of residual oil fly ash (ROFA) particle toxicity in rats by pulmonary hypertension and ambient temperature (Ta) change. Am J Respir Crit Care Med (1996) 153: A542

  10. Dockery, D, Pope, CA 3rd . Epidemiology of acute health effects: Summary of time-series studies. In: Wilson R, Spengler J, (Eds.)Particles in our air: Concentrations and health effects. Harvard University Press 1996 pp. 123–147

  11. Dockery, DW, Pope, ACD, Xu, X, Spengler, JD, Ware, JH, Fay, ME, Ferris, BG Jr., and Speizer, FE . An association between air pollution and mortality in six U.S. cities. N Engl J Med (1993) 329: 1753–1759

  12. Dyer, AR, Persky, V, Stamler, J, Paul, O, Shekelle, RB, Berkson, DM, Lepper, M, Schoenberger, JA, and Lindberg, HA . Heart rate as a prognostic factor for coronary heart disease and mortality: findings in three Chicago epidemiologic studies. Am J Epidemiol (1980) 112: 736–749

  13. Ebelt, ST, Fisher, TV, Petkau, AJ, Vedal, S, and Brauer, M . Exposure of chronic obstructive pulmonary disease (COPD) patients to particles: relationship between personal exposure and ambient air concentrations. J Air Waste Manage Assoc (2000) 50: 174–187

  14. Gillum, RF . The epidemiology of resting heart rate in a national sample of men and women: associations with hypertension, coronary heart disease, blood pressure, and other cardiovascular risk factors. Am Heart J (1988) 116: 163–174

  15. Godleski, JJ, Verrier, RL, Koutrakis, P, Catalano, P, Coull, B, Reinisch, U, Lovett, EG, Lawrence, J, Murthy, GG, Wolfson, JM, Clarke, RW, Nearing, BD, and Killingsworth, C . Mechanisms of morbidity and mortality from exposure to ambient air particles. Res Rep Health Effic Inst (2000 5–88discussion 89–103

  16. Gold, DR, Litonjua, A, Schwartz, J, Lovett, E, Larson, A, Nearing, B, Allen, G, Verrier, M, Cherry, R, and Verrier, R . Ambient pollution and heart rate variability. Circulation (2000) 101: 1267–1273

  17. Goldberg, RJ, Larson, M, and Levy, D . Factors associated with survival to 75 years of age in middle-aged men and women. Arch Intern Med (1996) 156: 505–509

  18. Gordon, T, Nadziejko, C, Chen, LC, and Schlesinger, R . Effects of concentrated ambient particles in rats and hamsters: an exploratory study [In Process Citation]. Res Rep Health Effic Inst 2000 1–34 discussion 35–42

  19. Gorecka, D . Cardiac arrhythmias in chronic obstructive pulmonary disease. Monaldi Arch Chest Dis (1997) 52: 278–281

  20. Hausberg, M, Mark, AL, Winniford, MD, Brown, RE, and Somers, VK . Sympathetic and vascular effects of short-term passive smoke exposure in healthy nonsmokers. Circulation (1997) 96: 282–287

  21. Jensen-Urstad, K, Storck, N, Bouvier, F, Ericson, M, Lindblad, LE, and Jensen-Urstad, M . Heart rate variability in healthy subjects is related to age and gender. Acta Physiol Scand (1997) 160: 235–241

  22. Kleiger, RE, Stein, PK, Bosner, MS, Rottman, JN . Time-domain measurements of heart variability. In: Malik M, Camm AJ, (Eds.) Heart Variability. Futura Publishing Company, Inc., Armonk, New York 1995 pp. 33–45

  23. Lewczuk, J, Wrabec, K, Piszko, P, Jagas, J, Sobkowicz-Wozniak, B, Porada, A, and Sawa, W . Chronic obstructive lung disease: does accompanying arrhythmia have any clinical significance? Kardiol Pol (1992) 37: 301–305

  24. Liao, D, Creason, J, Shy, C, Williams, R, Watts, R, and Zweidinger, R . Daily variation of particulate air pollution and poor cardiac autonomic control in the elderly. Environ Health Perspect (1999) 107: 521–525

  25. Linn, WS, Gong, H Jr., Clark, KW, and Anderson, KR . Day-to-day particulate exposures and health changes in Los Angeles area residents with severe COPD. Am J Respir Crit Care Med (1998) 157: A878

  26. NRC. Research priorities for airborne particulate matter: I. Immediate priorities and a long-range research portfolio: Committee on Research Priorities for Airborne Particulate Matter, National Research Council, 1998

  27. Penney, DG, Howley, JW . Is there a connection between carbon monoxide exposure and hypertension? Environ Health Perspect (1991) 95: 191–198

  28. Peters, A, Perz, S, Doring, A, Stieber, J, Koenig, W, and Wichmann, HE . Increases in heart rate during an air pollution episode. Am J Epidemiol (1999) 150: 1094–1098

  29. Pope, CA 3rd, and Kanner, RE . Acute effects of PM10 pollution on pulmonary function of smokers with mild to moderate chronic obstructive pulmonary disease. Am Rev Respir Dis (1993) 147: 1336–1340

  30. Pope, CA 3rd, Schwartz, J, and Ransom, MR . Daily mortality and PM10 pollution in Utah Valley. Arch Environ Health (1992) 47: 211–217

  31. Pope, CA 3rd, Bates, DV, and Raizenne, ME . Health effects of particulate air pollution: time for reassessment? Environ Health Perspect (1995) 103: 472–480

  32. Pope, CA 3rd, Thun, MJ, Namboodiri, MM, Dockery, DW, Evans, JS, Speizer, FE, and Heath, CW. Jr . Particulate air pollution as a predictor of mortality in a prospective study of U.S. adults. Am J Respir Crit Care Med (1995) 151: 669–674

  33. Pope, CA 3rd, Dockery, DW, Kanner, RE, Villegas, GM, and Schwartz, J . Oxygen saturation, pulse rate, and particulate air pollution: a daily time-series panel study. Am J Respir Crit Care Med (1999a) 159: 365–372

  34. Pope, CA 3rd, Verrier, RL, Lovett, EG, Larson, AC, Raizenne, ME, Kanner, RE, Schwartz, J, Villegas, GM, Gold, DR, and Dockery, DW . Heart rate variability associated with particulate air pollution. Am Heart J (1999b) 138: 890–899

  35. Schwartz, J. . Total suspended particulate matter and daily mortality in Cincinnati, Ohio. Environ Health Perspect (1994a) 102: 186–189

  36. Schwartz, J. . What are people dying of on high air pollution days? Environ Res (1994b) 64: 26–35

  37. Schwartz, J . Air pollution and hospital admissions for cardiovascular disease in Tucson. Epidemiology (1997) 8: 371–377

  38. Schwartz, J . Air pollution and hospital admissions for heart disease in eight U.S. counties. Epidemiology (1999) 10: 17–22

  39. Schwartz, J, and Dockery, DW . Increased mortality in Philadelphia associated with daily air pollution concentrations. Am Rev Respir Dis (1992) 145: 600–604

  40. Schwartz, J, Dockery, DW, and Neas, LM . Is daily mortality associated specifically with fine particles? J Air Waste Manage Assoc (1996) 46: 927–939

  41. Shih, HT, Webb, CR, Conway, WA, Peterson, E, Tilley, B, and Goldstein, S . Frequency and significance of cardiac arrhythmias in chronic obstructive lung disease. Chest (1988) 94: 44–48

  42. Singer, DH, Ori, Z . Changes in heart rate variability associated with sudden cardiac death. In: Malik M, Camm AJ, (Eds.) Heart Rate Variability. Futura Publishing Company, Inc., Armonk, New York 1995 pp. 429–448

  43. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Heart rate variability: standards of measurement, physiological interpretation, and clinical use. Eur Heart J (1996) 17: 354–381

  44. Thurston, GD . A critical review of PM10 mortality time-series studies. J Expos Anal Environ Epidemiol (1996) 6: 3–21

  45. Vedal, S . Ambient particles and health: lines that divide. J Air Waste Manage Assoc (1997) 47: 551–581

  46. Vybiral, T, Glaeser, DH . Changes of heart rate variability preceding ventricular arrhythmias. In: Malik M, Camm AJ, (Eds.) Heart Rate Variability. Futura Publishing Company, Inc., Armonk, New York 1995 pp. 421–428

  47. Watkinson, WP, Campen, MJ, and Costa, DL . Cardiac arrhythmia induction after exposure to residual oil fly ash particles in a rodent model of pulmonary hypertension. Toxicol Sci (1998) 41: 209–216

  48. Watkinson, WP, Campen, MJ, Nolan, JP, and Costa, DL . Cardiovascular and systemic responses to inhaled pollutants in rodents: effects of ozone and particulate matter. Environ Health Perspect (2001) 109 (Suppl 4) 539–546

Download references

Acknowledgements

We extend our sincere thanks to the study subjects who generously and conscientiously donated 7 days of personal sampling to this project. We would like to thank the Greater Vancouver Regional District Air Quality Department for their assistance with ambient monitoring, and Parveen Bhatti for technical assistance with the field monitoring and laboratory analysis.

Author information

Correspondence to MICHAEL BRAUER.

Additional information

Supported, in part, by grants from the British Columbia Lung Association, a British Columbia Lung Association/Medical Research Council of Canada Scientist (M. Brauer), and an American Lung Association Career Investigator Award (M. Brauer).

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • air pollution
  • environmental exposure
  • heart rate
  • heart rate variability
  • lung function
  • particles

Further reading