Skip Navigation


American Journal of Epidemiology Advance Access originally published online on June 22, 2006
American Journal of Epidemiology 2006 164(5):421-433; doi:10.1093/aje/kwj206
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
164/5/421    most recent
kwj206v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (7)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Symons, J. M.
Right arrow Articles by Geyh, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Symons, J. M.
Right arrow Articles by Geyh, A.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology Copyright © 2006 by the Johns Hopkins Bloomberg School of Public Health All rights reserved; printed in U.S.A.

Original Contribution

A Case-Crossover Study of Fine Particulate Matter Air Pollution and Onset of Congestive Heart Failure Symptom Exacerbation Leading to Hospitalization

J. M. Symons1, L. Wang1, E. Guallar1, E. Howell2, F. Dominici3, M. Schwab1, B. A. Ange1, J. Samet1, J. Ondov4, D. Harrison4 and A. Geyh5

1 Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD
2 Johns Hopkins Bayview Medical Center, Baltimore, MD
3 Department of Biostatistics, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD
4 Department of Chemistry and Biochemistry, College of Chemical and Life Sciences, University of Maryland, College Park, MD
5 Department of Environmental Health Sciences, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, MD

Correspondence to Dr. Alison Geyh, Department of Environmental Health Sciences, Johns Hopkins Bloomberg School of Public Health, 615 N. Wolfe Street, Room E6654, Baltimore, MD 21205 (e-mail: ageyh{at}jhsph.edu).

Received for publication August 15, 2005. Accepted for publication January 26, 2006.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Persons with congestive heart failure may be susceptible to ambient air pollution. The authors evaluated the association between exposure to particulate matter with an aerodynamic diameter of <2.5 µm (PM2.5) and onset of symptom exacerbation leading to hospital admission in Baltimore, Maryland. They used a case-crossover design for 135 case events occurring among 125 persons with prevalent congestive heart failure who were admitted to a single hospital through the emergency department during 2002. The case period was assigned using three index times: 8-hour and 24-hour periods of symptom onset and date of hospital admission. Controlling for weather, the authors detected a modest relative increase in risk for cases defined by 8-hour symptom onset for an interquartile-range increase in PM2.5 at a 2-day lag (odds ratio = 1.09, 95% confidence interval: 0.91, 1.30). A corresponding increase in risk was not observed when admission date was used to define the case period. A series of simulations based on study data indicated that the study had adequate statistical power to detect odds ratios of 1.2 or higher. Although overall findings were not statistically significant, the identification of case events defined by an 8-hour onset period may be more relevant than either a 24-hour onset period or the admission date for estimating harmful effects of air pollutant exposure on cardiovascular health.

air pollutants; disease susceptibility; dyspnea; environmental exposure; heart failure, congestive; hospitalization

Abbreviations: BMCA, Bayview Medical Center admission date; CHF, congestive heart failure; CI, confidence interval; PM2.5, particulate matter with an aerodynamic diameter of <2.5 µm; PM10, particulate matter with an aerodynamic diameter of <10 µm; TOS, time of onset for symptom exacerbation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Fine-particle air pollution, defined as particulate matter with an aerodynamic diameter of <2.5 µm (PM2.5), is associated with increased all-cause and cause-specific mortality (1Go, 2Go) and with increased respiratory and cardiovascular morbidity (3Go–5Go). While mechanisms through which inhaled particulate matter injures the pulmonary system have been documented (6Go, 7Go), understanding of the biologic processes by which particulate matter may affect the cardiovascular system remains incomplete (8Go). Epidemiologic research has identified adverse physiologic effects associated with increased particulate matter exposure in persons with cardiovascular disease, including reduced heart rate variability (9Go, 10Go), increased blood pressure (11Go), cardiac arrhythmias (12Go), increased oxidative stress and inflammation (13Go), and progression of atherosclerosis (14Go, 15Go).

Among persons with cardiovascular disease, those with congestive heart failure (CHF) may be especially sensitive to ambient air pollutants, principally PM2.5 (16Go, 17Go). In general, the rising incidence of chronic heart failure has led to an increased number of persons who are potentially susceptible to adverse health effects associated with particulate matter (18Go). CHF, a clinical syndrome resulting from pump failure of the cardiac muscle (19Go), is characterized by symptoms that include shortness of breath, fatigue, and edema resulting in weight gain and swelling of lower body extremities. As heart failure worsens, patients typically experience acute, severe symptom exacerbations that require medical care, usually through a hospital emergency department (20Go). Worsening CHF is responsible for over one million hospital admissions annually, representing one of the largest categories of annual Medicare expenditures (21Go).

The present pilot investigation focuses on the onset of dyspnea as a cardinal indicator of CHF decompensation; its timing should be more directly related to particulate matter exposure than hospital admission date. In this analysis, we sought to determine the relation between ambient PM2.5 exposure and development of symptoms sufficiently severe to lead to hospital admission through the emergency department. Specifically, we identified onset times for exacerbation of heart failure symptoms in relation to subsequent emergency hospital admission; evaluated the association between onset time for exacerbation of heart failure symptoms and ambient particulate matter exposure levels; and assessed the ability of the case-crossover analytic method to detect specific effect estimates using simulated case data and empirical particulate matter measurements.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
Recruitment of the study population
Study participants were patients diagnosed with CHF who were admitted to Johns Hopkins Bayview Medical Center in Baltimore, Maryland, through the emergency department. Bayview Medical Center was selected because it was located adjacent to the Baltimore Particulate Matter Supersite, an air quality monitoring station that conducted intensive monitoring of PM2.5 (22Go, 23Go). Participant recruitment occurred from April to December of 2002, coinciding with the data collection period of the Baltimore Supersite. Eligible participants were identified through daily review of emergency department admission logs by a hospitalist who examined patient charts to verify the diagnosis of CHF. Trained interviewers screened potentially eligible patients and obtained informed consent for participation prior to conducting an in-hospital interview. Patients were excluded at the interview stage if they were cognitively impaired, could not verbally communicate, or could not speak English. The study was approved by the institutional review board of the Johns Hopkins Bloomberg School of Public Health.

Data collection
Hospitalized persons were interviewed after admission through the emergency department during their stays in overnight wards of the hospital. All participants responded to an interviewer-administered questionnaire that covered awareness and self-evaluation of symptoms occurring prior to admission. The questionnaire was designed to collect relevant information regarding symptom onset, health conditions, and factors related to air pollution exposure. Interviews conducted using the questionnaire were intended to take no more than 20 minutes, in order to minimize time demands on participants during their hospitalization. Participants were asked to assess the onset of exacerbation of heart failure decompensation by identifying the day on which "breathing became more difficult than usual" and their symptoms necessitated emergency medical attention. This time of onset for symptom exacerbation (TOS) was used as the decompensation event time for the period associated with particulate matter exposure and risk of emergency admission. If possible, participants further specified an 8-hour period of the day for shortness-of-breath onset.

Air pollutant and weather measurement
PM2.5 mass concentration and copollutant data, including levels of nitrogen dioxide and carbon monoxide, were provided by researchers at the Baltimore Supersite (22Go, 23Go). Additional data recorded at monitoring sites located near the recruitment hospital, including ozone and meteorologic measurements, were obtained from the Maryland Department of the Environment. All air pollutant measurements were reported for 1-hour intervals. All weather data were measured at 5-minute intervals and averaged for 8-hour and 24-hour periods.

The main exposure metric used for this analysis was ambient PM2.5 mass concentration. The original study design assumed that all PM2.5 data would be provided by researchers at the Baltimore Supersite; however, because of an instrument malfunction, data were unavailable from May 23, 2002, to July 4, 2002—approximately 17 percent of the study period. Subsequently, we obtained mass concentration data from seven other monitoring stations maintained by the Maryland Department of the Environment in Baltimore City to estimate the PM2.5 series for the study period. Table 1 lists these eight monitoring stations, as well as measurement methods, sampling intervals, and pairwise correlation coefficients for the PM2.5 concentrations.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Locations of PM2.5* monitoring stations, sampling methods, and correlation coefficients for collection of data on 24-hour average PM2.5 concentrations, Baltimore, Maryland, 2002

 
To estimate average PM2.5 concentrations, we fitted a generalized linear model with an autoregressive measurement error term of the first order, AR(1Go) (24Go). The estimation model is expressed as Ztj = {Theta}t + {varepsilon}tj, {varepsilon}tj ~ AR(1Go), where Ztj is the PM2.5 mass concentration at time t measured at monitoring station j, {Theta}t is the "true" PM2.5 mass concentration at time t, and {varepsilon}tj is the measurement error at time t at monitoring station j. Inputs for daily estimates came from all monitoring stations, while inputs for hourly estimates were obtained from only the Baltimore Supersite and the Old Town site, using tapered element oscillating microbalances (Rupprecht and Pataschnick, Albany, New York). A larger estimated standard error was applied to hourly concentration estimates for the time period with missing Baltimore Supersite data. Hourly estimates of PM2.5 concentrations were averaged to obtain an 8-hour exposure measurement.

Study design
A case-crossover design was used to assess the risk of CHF symptom exacerbation and ensuing emergency admission in relation to ambient PM2.5 levels. In this design, a specified referent window is identified for each subject that includes the case event matched to a set of control, or nonevent, periods for the same individual. The analytic approach compares exposures occurring during case events, defined as the "hazard" or "at-risk" period, with exposures for control periods, in which the outcome of interest did not occur (25Go). By making within-person comparisons, the case-crossover design eliminates confounding by fixed personal characteristics, such as age and gender (26Go). For this study, selection of the case period was assigned to two index dates: the participant-identified day of symptom onset (TOS) and the day of emergency department admission (Bayview Medical Center admission date (BMCA)). To control for time-varying factors and remove potential overlap bias, we used a time-stratified referent selection approach (27Go–29Go). We matched control periods by day of the week in the same calendar month as the case period in order to account for potential time trends in the exposure series (30Go, 31Go). For 8-hour averaged exposures, case and control periods were assigned to one of three 8-hour onset periods (morning = 4 a.m. to <12 p.m.; afternoon = 12 p.m. to <8 p.m.; night = 8 p.m. to <4 a.m.). If the participant was unable to identify an 8-hour period of the day for TOS, we assigned the period using a random process with the probabilities of morning (40 percent), afternoon (30 percent), and evening (30 percent) onset based on the distribution of times for participant-identified TOS periods.

Statistical methods
To assess the relation between PM2.5 exposure and defined case events, we fitted a conditional logistic regression model that estimated the relative risk associated with interquartile-range increases in exposure to ambient PM2.5 and gaseous pollutants for 8-hour and 24-hour averaged mass concentrations. Models were fitted for three case periods: 1) the participant's onset period (TOS8h), 2) the participant's onset date (TOS24h), and 3) BMCA. Since the induction time for symptom exacerbation related to fine particulate matter exposure was assumed to be limited to an acute period, we investigated single and cumulative lagged exposure periods ranging from 0 days to 3 days (32Go). Analyses were performed for both single-pollutant models (PM2.5, carbon monoxide, nitrogen dioxide, and ozone) and two-pollutant models (PM2.5 with a copollutant), unadjusted and adjusted for temperature and humidity (33Go, 34Go). All analyses were performed using the statistical software packages SAS (version 8.1; SAS Institute, Inc., Cary, North Carolina) and S-PLUS (version 6.1; Insightful Corporation, Seattle, Washington).

Simulation study
To assess the lower bound of detectable effects using this study design, we conducted a simulation study with 135 case events distributed over the 244-day study period. For the conditional logistic regression model, the probability that an event occurs at time t is taken from the proportional hazards model (28Go). The resulting probability function after removal of the baseline hazard rate used to randomly assign case events is Formula where Xt represents the series of daily average pollutant levels measured on day t and the coefficient ß represents the known or "true" effect estimate. We applied the observed 24-hour average PM2.5 measurements to three sets of simulations for ß equal to 0.01, 0.02, and 0.05, with 500 data sets being randomly generated for each fixed estimate (35Go, 36Go). These values were chosen to be consistent with odds ratio estimates for interquartile-range differences in 24-hour PM2.5 of 1.10, 1.20, and 1.58, respectively. Simulated case data sets were analyzed using the same conditional logistic regression model, adjusted for temperature and humidity, as that used in the empirical analysis with no exposure lag period.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
From April through November of 2002, we identified 398 admissions of patients to Bayview Medical Center for overnight stays following CHF symptom exacerbation. Overall, 127 persons completed 137 interviews conducted a median of 2 days after admission (range, 0–16 days). In addition to the exclusions listed above (n = 62), some patients were not interviewed because of lack of consent (n = 86) or our inability to contact them during their hospitalization (e.g., they were not in the room during interviewer visits; n = 113). Two persons with primary residences outside the state of Maryland were excluded. Therefore, the final sample comprised 125 persons, of whom eight were interviewed twice and one was interviewed three times. This resulted in a total of 135 case events of hospital admission due to exacerbation of CHF symptoms. All repeat cases occurred during separate calendar months of the study period; therefore, they were assumed to be independent events for this analysis (35Go). Figure 1 shows the geographic distribution of participants in the study area defined around the primary air pollutant monitoring site, the Baltimore Supersite. Addresses for 101 participants are plotted, while 24 participants with missing information on street address were enumerated by residential ZIP code only.


Figure 1
View larger version (24K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 1. Residence locations for participants (n = 101) in a study of exacerbation of congestive heart failure symptoms by fine particulate air pollution, Baltimore, Maryland, 2002. Participants without street address information (n = 23) are enumerated in six residential ZIP codes, indicated by black circles. One participant residing in ZIP code 20724 is not shown on the figure.

 
The median age of study participants was 70 years. Participants were more likely to be female (61 percent) and White (79 percent) than nonparticipants (table 2). Although few participants were current smokers (14 percent), almost 40 percent had experienced passive exposure to tobacco smoke in their homes. A substantial proportion of participants had indications of comorbid disease in their medical charts, including high levels of hypertension (68 percent), diabetes (54 percent), and coronary heart disease (50 percent), as well as other chronic cardiovascular and respiratory conditions.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Characteristics of participants in a study of exacerbation of congestive heart failure symptoms by fine particulate air pollution (n = 125), Baltimore, Maryland, 2002

 
TOS for shortness of breath
Figure 2 displays case periods based on the identified day of TOS with further assignment of the 8-hour period for onset prior to admission day. Participants varied in interval from symptom exacerbation onset to emergency admission, with most participants (88 percent) visiting the emergency department within 1 day of TOS (range, 0–7 days). Participants identified an 8-hour period of the day for symptom exacerbation (TOS8h) in 97 interviews. For 38 cases without an 8-hour TOS time, we assigned this period using a random process based on proportions of identified 8-hour TOS periods. Overall, 56 case events (41 percent) had morning onset, 39 (29 percent) had afternoon onset, and 40 (30 percent) had evening onset.


Figure 2
View larger version (11K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 2. Distribution of participant-identified 8-hour periods of onset of congestive heart failure symptom exacerbation preceding hospital admission (135 admissions), Baltimore, Maryland, 2002. M, morning period (4 a.m. to <12 p.m.); A, afternoon period (12 p.m. to <8 p.m.); E, evening period (8 p.m. to <4 a.m.).

 
Summary of exposure measures
Figure 3 illustrates the temporal series of daily measures for each monitoring site and the averaged time-series results for daily PM2.5 concentrations as estimated by the AR(1Go) model. Each series represents the 24-hour averaged concentrations for the eight monitoring stations listed in table 1. Measurements of PM2.5 provided evidence of a strong correlation between monitoring stations (Pearson's r > 0.9 for all two-way comparisons). The overall time series at the bottom of the figure represents the estimated PM2.5 daily average for Baltimore City. Sensitivity analyses found that the generalized linear model results were robust to the specification of more complex autoregressive structures than AR(1Go).


Figure 3
View larger version (11K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 3. Time-series estimates of daily average and overall concentrations (µg/m3) of particulate matter with an aerodynamic diameter of <2.5 µm from eight monitoring stations, Baltimore, Maryland, 2002. FRM, federal reference method; TEOM, tapered element oscillating microbalance.

 
Table 3 summarizes the distributions of 8-hour and 24-hour averaged mean values for air pollutants and weather variables. Ambient PM2.5 was moderately correlated with nitrogen dioxide (r = 0.53), ozone (r = 0.48), and temperature (r = 0.51) and was weakly correlated with carbon monoxide (r = 0.18) and relative humidity (r = –0.08). Table 4 shows the distribution of differences in PM2.5 concentrations between case and control periods by lag time (37Go). For referent windows using 8-hour averaged periods, exposure concentrations for control periods are higher than those for case periods at a lag time of 0, while at a lag of 2 days, average case period exposures differ by 2.2 µg/m3 from corresponding control periods. For each case definition and lag period, 75 percent of referent windows differ by less than the interquartile range of time-averaged PM2.5 for the study period.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Exposure variables included in a study of exacerbation of congestive heart failure symptoms by fine particulate air pollution, by averaging period, Baltimore, Maryland, 2002

 

View this table:
[in this window]
[in a new window]

  		TABLE 4. Differences between case and control PM2.5* exposure concentrations by case event referent window, Baltimore, Maryland, 2002

 
Analytic results
Table 5 presents estimates of the odds ratio from single-pollutant analyses. Generally, point estimates were near 1.0, with 95 percent confidence intervals extending below and above this null value. For single-lag analyses, the odds ratios for PM2.5 for TOS8h increased from a lag of 0 to a lag of 3 days. Only a single association for TOS8h reached statistical significance (p < 0.05): that for carbon monoxide at a 2-day lag (odds ratio = 1.23, 95 percent confidence interval (CI): 1.01, 1.51). For both 24-hour averaged outcomes, odds ratio estimates were consistently below the null value of 1.0 in single-pollutant models for PM2.5, carbon monoxide, and nitrogen dioxide. In single-pollutant models for ozone, odds ratio estimates above unity were observed for TOS24h in association with lagged exposures at 2 and 3 days and with BMCA at a lag of 1 day.


View this table:
[in this window]
[in a new window]

  		TABLE 5. Odds ratio estimates for defined case events associated with an interquartile-range increase in pollutant concentrations in single-pollutant models, Baltimore, Maryland, 2002*

 
We also assessed the effect of PM2.5 with adjustment for other pollutants in two-pollutant models. Similar to the findings from single-pollutant models, PM2.5 risk estimates were generally below the null value and did not attain statistical significance. A trend of increased odds ratios with lengthening lag was observed for TOS8h after adjustment for copollutants and weather variables. Risk estimates for this outcome, though close to the null, were maximized at 2-day exposure lags for each set of model estimates (adjusted for ozone, odds ratio = 1.09, 95 percent CI: 0.91, 1.31 (displayed in figure 4); adjusted for carbon monoxide, odds ratio = 1.03, 95 percent CI: 0.86, 1.25 (results not shown); and adjusted for nitrogen dioxide, odds ratio = 1.04, 95 percent CI: 0.85, 1.27 (results not shown)). A similar pattern was not observed for TOS24h or BMCA case events with adjustment for gaseous copollutants and weather variables (figure 4).


Figure 4
View larger version (10K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 4. Adjusted odds ratios for the relation between exposure to particulate matter with an aerodynamic diameter of <2.5 µm and exacerbation of congestive heart failure symptoms leading to hospitalization, by lag period (days), for three case definitions (8-hour time of onset for symptom exacerbation (TOS8h), 24-hour time of onset for symptom exacerbation (TOS24h), and Bayview Medical Center admission date (BMCA)), Baltimore, Maryland, 2002. Results were adjusted for ozone level, temperature, and humidity in all models. Bars, 95% confidence interval.

 
Simulation results
Figure 5 illustrates the distribution of regression coefficient estimates (ß) provided by the conditional logistic regression model for 500 iterations of data simulated for the three fixed ß values. The results indicated that our study design was sufficiently powerful to detect statistically significant effects for odds ratios greater than 1.20 for an interquartile-range difference in 24-hour averaged PM2.5. However, uncertainty increased for estimates when we tried to detect a smaller fixed effect of ß equal to 0.01 that was consistent with an odds ratio of 1.10.


Figure 5
View larger version (38K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 5. Estimated results from 500 simulations each for three fixed values of ß (beta coefficient from conditional logistic regression model).

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
Our study provides new insights into the temporal dynamics of symptom onset in persons with CHF that may be useful for future research. Although the primary results did not demonstrate a statistically significant association between case events and exposure to PM2.5, a trend of increased risk estimates at lengthening lagged exposures was observed when we used the 8-hour symptom onset period to define case events. This relation was not seen when 24-hour exposure periods were evaluated using either onset day or hospital admission date to define case events.

In this case-crossover study, we found that a more precise definition of disease and exposure timing allowed for the potential detection of a more specific association between exposure to air pollutants, specifically PM2.5, and exacerbation of CHF symptoms. We developed an interview approach that described the timing of CHF decompensation and were able to explore the impact of PM2.5 exposure for clinically important periods prior to hospitalization. We hypothesized that this approach would enhance detection of a more relevant exposure period for PM2.5 associated with cardiovascular effects. The use of a case-crossover design allowed for the control of confounding due to time-fixed characteristics for persons with CHF (38Go), and the choice of time-stratified referent sampling minimized bias due to seasonal and weekly trends in PM2.5 levels (27Go, 29Go).

Previous studies have explored the relation between CHF morbidity and exposure to air pollutants (17Go, 39Go–44Go). The majority of these studies have demonstrated positive and significant associations with at least one of the pollutants examined. For example, in a study carried out in Detroit, Michigan, Schwartz and Morris (42Go) found a significant association between heart failure hospital admission and concentrations of carbon monoxide and particulate matter with an aerodynamic diameter of <10 µm (PM10) (for carbon monoxide, relative risk = 1.02, 95 percent CI: 1.01, 1.03; for PM10, relative risk = 1.02, 95 percent CI: 1.00, 1.04). In a study carried out in Denver, Colorado, Koken et al. (43Go) verified a significant association with CHF admission only for carbon monoxide at a 3-day lag (relative risk = 1.10, 95 percent CI: 1.00, 1.22). In contrast, in a study carried out in London, United Kingdom, Poloniecki et al. (44Go) found no statistically significant relation between any pollutant and hospital admission for heart failure.

In most of these studies, the primary outcome was hospital admission, with disease status being determined by International Classification of Diseases, Ninth Revision, codes. Previous research has indicated that reliance on International Classification of Diseases, Ninth Revision, codes may result in undercounting of hospitalizations of persons with clinical evidence of CHF by as much as 33 percent (45Go). The potential miscoding or exclusion of significant numbers of CHF patients in large studies where disease is identified solely through discharge codes may result in outcome misclassification arising from hospital administrative decisions (46Go). In addition, exposure in the referenced studies was estimated on the basis of lag structures defined by time of admission, which may not be biologically relevant to onset of symptom exacerbation.

We did find carbon monoxide to be significantly associated with onset of heart failure symptom exacerbation for 8-hour exposures at a 2-day lag. Our positive finding for this pollutant is consistent with other studies of heart failure morbidity and air pollution exposure. In these studies, carbon monoxide, of all the pollutants considered, has been most consistently found to be related to CHF (17Go, 39Go–44Go). Carbon monoxide reduces oxygen delivery and has been linked to myocardial ischemia at low concentrations (47Go, 48Go).

We also observed maximum effect estimates at a 2-day lagged exposure in analyses evaluating participant-identified 8-hour TOS as the case period. This result is similar to that of studies which have detected associations for elevated levels of fine particulate matter 2 days prior to specific adverse cardiovascular events with identifiable acute onset times, such as primary cardiac arrest (49Go) and automated implantable cardiac defibrillator firings (12Go). However, our findings differ from those of large time-series studies in which investigators reported stronger associations between CHF admission and PM2.5 level on the day of hospital admission (50Go).

One explanation for this discrepancy may be differences between outcomes defined for individual-based case-crossover studies and those identified from larger admissions databases. In this study, each participant was recruited on the basis of a CHF diagnosis confirmed by a physician, using information obtained from hospital admission charts. This approach may be more specific for case-finding in comparison with studies that rely on discharge diagnoses coded by hospital billing departments. Moreover, persons with CHF are a clinically heterogeneous group with a high proportion of comorbid diseases, such as chronic obstructive pulmonary disease, hypertension, and diabetes, that may be potential sources of outcome misclassification (51Go). High rates of these conditions were observed among participants in our study, illustrating the complicated nature of medical classification of persons with CHF.

There are several potential limitations to our findings. As with other studies of ambient air pollution, exposure measurement error may have led us to incorrectly estimate the true pollutant exposures for study participants (52Go). The use of multiple monitors to complete the measurement series of PM2.5 demonstrated a high level of homogeneity for ambient PM2.5 concentrations estimated throughout the study area. Previous research has demonstrated that personal and ambient concentrations of PM2.5 are highly correlated in populations similar to the one in this study (53Go, 54Go). Additionally, some participants, mostly at the beginning of the study, were unable to specify an 8-hour period for TOS. During the course of the study, refinements in the interview allowed us to improve patients' identification of TOS for 8-hour onset periods. A comparative analysis found little bias in using this random assignment process. Finally, time-varying risks such as individual compliance with physician recommendations concerning diet, activity, and medications were not assessed because of concerns about the length of the interview. These factors may exert an acute effect on the ability to assess hospitalization risk for persons with CHF either independently of or as modified by elevated fine particulate matter exposures (55Go).

In planning this study, we recognized that statistical power might be limited by the sample size, that is, the number of persons who could be feasibly recruited. Our empirical analyses detected an increasing trend in risk but lacked statistical significance. To better interpret our findings, we conducted three simulation analyses using 135 case events assigned to the observed daily PM2.5 concentrations for our study period. The results indicated that if the "true" odds ratio were 1.20 or higher for an interquartile-range difference in observed PM2.5 ambient concentration, our study design was sufficiently powerful to detect this level of increased risk.

The generally null findings of our empirical analysis weigh against strong effect estimates for risk of symptom exacerbation due to PM2.5 exposures. The results of the simulations conducted for a fixed ß equal to 0.01 (equivalent to an odds ratio of 1.10) indicate that our design was able to detect this smaller effect, though many individual model estimates lacked statistical significance. This simulated value is similar to our maximum estimate of association between 8-hour averaged ambient PM2.5 and symptom onset detected at an exposure lag of 2 days. Findings from the simulation study suggest that the absence of statistical evidence does not necessarily represent evidence for the absence of an effect. Rather, it is likely that the small magnitude of PM2.5 effects requires a larger number of cases in order to obtain sufficient power for detection using this design. Furthermore, comparison of exposure differences between case and control periods indicates that our null findings may also have resulted from relatively small absolute differences in exposure concentrations within referent windows (37Go).

If PM2.5 has a causal effect in worsening the health of persons with chronic illnesses such as CHF, a potentially large number of excess hospitalizations can be attributed to this ubiquitous exposure (56Go). Exacerbations of CHF symptoms have been associated with increased ambient PM2.5 levels; however, individual variation in the severity and timing of responses may be due to differing physiologic mechanisms (57Go). Evidence from studies of shortness of breath and other heart failure decompensation symptoms demonstrates that exacerbations may progress over a period of several days following a triggering exposure (58Go, 59Go). Assigning case periods by individual onset time may provide a useful model for estimating the risk of hospitalization associated with increased levels of ambient PM2.5. Although our findings are inconclusive, they suggest that exposures sufficient to affect health may be occurring earlier than would be inferred from studies based only on date of hospital admission. The understanding of induction time is a key element for determining which biologic mechanisms that adversely affect the health of susceptible persons are influenced by fine particulate matter exposure.


   ACKNOWLEDGMENTS
 
Support for this study was provided by the Electric Power Research Institute (Palo Alto, California) under grant EP-P7198/C3654. Additional support came from the American Petroleum Institute (Washington, DC). This research was conducted in complement with the Baltimore Supersite Project, which was supported by the US Environmental Protection Agency (contract R82806101).

The authors thank Ajay Gupta for data collection, D'Ann Williams for mapping, Sorina Eftim for statistical modeling, and Yun Lu and Dr. Aidan McDermott for help with the simulation study.

Conflict of interest: none declared.


   NOTES
 
Editor's note: An invited commentary on this article appears on page 434.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Dockery DW, Pope CA III, Xu X, et al. An association between air pollution and mortality in six U.S. cities. N Engl J Med 1993;329:1753–9.[Abstract/Free Full Text]
  2. Pope CA III, Burnett RT, Thun MJ, et al. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA 2002;287:1132–41.[Abstract/Free Full Text]
  3. Morris RD. Airborne particulates and hospital admissions for cardiovascular disease: a quantitative review of the evidence. Environ Health Perspect 2001;109(suppl 4):495–500.
  4. Peel JL, Tolbert PE, Klein M, et al. Ambient air pollution and respiratory emergency department visits. Epidemiology 2005;16:164–74.[CrossRef][Web of Science][Medline]
  5. Verrier RL, Mittleman MA, Stone PH. Air pollution: an insidious and pervasive component of cardiac risk. Circulation 2002;106:890–2.[Free Full Text]
  6. MacNee W, Donaldson K. Particulate air pollution: injurious and protective mechanisms in the lungs. In: Holgate ST, Koren HS, Samet JM, et al, eds. Air pollution and health. San Diego, CA: Academic Press, 1999:653–72.
  7. Kelly FJ. Oxidative stress: its role in air pollution and adverse health effects. Occup Environ Med 2003;60:612–16.[Free Full Text]
  8. Brook RD, Franklin B, Cascio W, et al. Air pollution and cardiovascular disease: a statement for healthcare professionals from the Expert Panel on Population and Prevention Science of the American Heart Association. Circulation 2004;109:2655–71.[Abstract/Free Full Text]
  9. Gold DR, Litonjua A, Schwartz J, et al. Ambient pollution and heart rate variability. Circulation 2000;101:1267–73.[Abstract/Free Full Text]
  10. Devlin RB, Ghio AJ, Kehrl H, et al. Elderly humans exposed to concentrated air pollution particles have decreased heart rate variability. Eur Respir J Suppl 2003;40:76s–80s.[CrossRef][Medline]
  11. Ibald-Mulli A, Stieber J, Wichmann HE, et al. Effects of air pollution on blood pressure: a population-based approach. Am J Public Health 2001;91:571–7.[Abstract]
  12. Peters A, Liu E, Verrier RL, et al. Air pollution and incidence of cardiac arrhythmia. Epidemiology 2000;11:11–17.[CrossRef][Web of Science][Medline]
  13. Donaldson K, Stone V, Seaton A, et al. Ambient particle inhalation and the cardiovascular system: potential mechanisms. Environ Health Perspect 2001;109(suppl 4):523–7.
  14. Suwa T, Hogg JC, Quinlan KB, et al. Particulate air pollution induces progression of atherosclerosis. J Am Coll Cardiol 2002;39:935–42.[Abstract/Free Full Text]
  15. Kunzli N, Jerrett M, Mack WJ, et al. Ambient air pollution and atherosclerosis in Los Angeles. Environ Health Perspect 2005;113:201–6.[Web of Science][Medline]
  16. Goldberg MS, Burnett RT, Bailar JC III, et al. Identification of persons with cardiorespiratory conditions who are at risk of dying from the acute effects of ambient air particles. Environ Health Perspect 2001;109(suppl 4):487–94.
  17. Wellenius G, Bateson TF, Mittleman M, et al. Particulate air pollution and the rate of hospitalization for congestive heart failure among Medicare beneficiaries in Pittsburgh, Pennsylvania. Am J Epidemiol 2005;161:1030–6.[Abstract/Free Full Text]
  18. Redfield MM. Heart failure—an epidemic of uncertain proportions. N Engl J Med 2002;347:1442–4.[Free Full Text]
  19. Francis GS, Tang WH. Pathophysiology of congestive heart failure. Rev Cardiovasc Med 2003;4(suppl 2):S14–20.
  20. Evangelista LS, Dracup K, Doering LV. Treatment-seeking delays in heart failure patients. J Heart Lung Transplant 2000;19:932–8.[CrossRef][Web of Science][Medline]
  21. Ghali JK. Decompensated heart failure revisited. (Editorial). Am J Med 2003;114:695–6.[CrossRef][Web of Science][Medline]
  22. Ondov JM, Buckley TJ, Hopke PK, et al. Baltimore Supersite: highly time and size resolved concentrations of urban PM2.5 and its constituents for resolution of sources and immune responses. Revision 1, December 15, 1999. (Project proposal). College Park, MD: Department of Chemistry and Biochemistry, University of Maryland, 1999. (http://www.chem.umd.edu/supersite/project/proposal-full.pdf).
  23. Ambient Monitoring Technology Information Center, Environmental Protection Agency. PM Supersites information. Washington, DC: Environmental Protection Agency, 2000. (http://www.epa.gov/ttn/amtic/supersites.html).
  24. Diggle PJ, Heagerty P, Liang KY, et al. Analysis of longitudinal data. Oxford, United Kingdom: Oxford University Press, 2002.
  25. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 1991;133:144–53.[Abstract/Free Full Text]
  26. Dominici F, Sheppard L, Clyde M. Health effects of air pollution: a statistical review. Int Stat Rev 2003;71:243–76.
  27. Lumley T, Levy D. Bias in the case-crossover design: implications for studies of air pollution. Environmetrics 2000;11:689–704.[CrossRef][Web of Science]
  28. Levy D, Lumley T, Sheppard L, et al. Referent selection in case-crossover analyses of acute health effects of air pollution. Epidemiology 2001;12:186–92.[CrossRef][Web of Science][Medline]
  29. Janes H, Sheppard L, Lumley T. Overlap bias in the case-crossover design, with application to air pollution exposures. Stat Med 2005;24:285–300.[CrossRef][Web of Science][Medline]
  30. Greenland S. Confounding and exposure trends in case-crossover and case-time-control designs. Epidemiology 1996;7:231–9.[Web of Science][Medline]
  31. Bateson TF, Schwartz J. Control for seasonal variation and time trend in case-crossover studies of acute effects of environmental exposures. Epidemiology 1999;10:539–44.[CrossRef][Web of Science][Medline]
  32. Schwartz J. The distributed lag between air pollution and daily deaths. Epidemiology 2000;11:320–6.[CrossRef][Web of Science][Medline]
  33. Braga AL, Zanobetti A, Schwartz J. The effect of weather on respiratory and cardiovascular deaths in 12 U.S. cities. Environ Health Perspect 2002;110:859–63.
  34. Schwartz J, Samet JM, Patz JA. Hospital admissions for heart disease: the effects of temperature and humidity. Epidemiology 2004;15:755–61.[CrossRef][Web of Science][Medline]
  35. Checkoway H, Levy D, Sheppard, L, et al. A case-crossover analysis of fine particulate matter and out-of-hospital cardiac arrest. (Health Effects Institute report no. 99). Cambridge, MA: Health Effects Institute, 2000.
  36. Figueiras A, Caracedo-Martinez E, Saez M, et al. Analysis of case-crossover designs using longitudinal approaches: a simulation study. Epidemiology 2005;16:239–46.[CrossRef][Web of Science][Medline]
  37. Kunzli N, Schindler C. A call for reporting the relevant exposure term in air pollution case-crossover studies. J Epidemiol Community Health 2005;59:527–30.[Abstract/Free Full Text]
  38. Kwon HJ, Cho SH, Nyberg F, et al. Effects of ambient air pollution on daily mortality in a cohort of patients with congestive heart failure. Epidemiology 2001;12:413–19.[CrossRef][Web of Science][Medline]
  39. Morris RD, Naumova EN, Munasinghe RL. Ambient air pollution and hospitalization for congestive heart failure among elderly people in seven large US cities. Am J Public Health 1995;85:1361–5.[Abstract/Free Full Text]
  40. Burnett RT, Dales RE, Brook JR, et al. Association between ambient carbon monoxide levels and hospitalizations for congestive heart failure in the elderly in 10 Canadian cities. Epidemiology 1997;8:162–7.[Web of Science][Medline]
  41. Morris RD, Naumova EN. Carbon monoxide and hospital admissions for congestive heart failure: evidence of an increased effect at low temperatures. Environ Health Perspect 1998;106:649–53.[Web of Science][Medline]
  42. Schwartz J, Morris R. Air pollution and hospital admissions for cardiovascular disease in Detroit, Michigan. Am J Epidemiol 1995;142:23–35.[Abstract/Free Full Text]
  43. Koken PJ, Piver WT, Ye F, et al. Temperature, air pollution, and hospitalization for cardiovascular diseases among elderly people in Denver. Environ Health Perspect 2003;111:1312–17.[Web of Science][Medline]
  44. Poloniecki JD, Atkinson RW, de Leon AP, et al. Daily time series for cardiovascular hospital admissions and previous day's air pollution in London, UK. Occup Environ Med 1997;54:535–40.[Abstract/Free Full Text]
  45. Goff DC, Pandey DK, Chan FA, et al. Congestive heart failure in the United States: is there more than meets the I(CD code)? The Corpus Christi Heart Project. Arch Intern Med 2000;160:197–202.[Abstract/Free Full Text]
  46. Fisher ES, Whaley FS, Krushat WM, et al. The accuracy of Medicare's hospital claims data: progress has been made, but problems remain. Am J Public Health 1992;82:243–8.[Abstract/Free Full Text]
  47. Allred EN, Bleecker ER, Chaitman BR, et al. Effects of carbon monoxide on myocardial ischemia. Environ Health Perspect 1991;91:89–132.[Web of Science][Medline]
  48. Townsend CL, Maynard RL. Effects on health of prolonged exposure to low concentrations of carbon monoxide. Occup Environ Med 2002;59:708–11.[Free Full Text]
  49. Sullivan J, Ishikawa N, Sheppard L, et al. Exposure to ambient fine particulate matter and primary cardiac arrest among persons with and without clinically recognized heart disease. Am J Epidemiol 2003;157:501–9.[Abstract/Free Full Text]
  50. Metzger KB, Tolbert PE, Klein M, et al. Ambient air pollution and cardiovascular emergency department visits. Epidemiology 2004;15:46–56.[CrossRef][Web of Science][Medline]
  51. Havranek EP, Masoudi FA, Westfall KA, et al. Spectrum of heart failure in older patients: results from the National Heart Failure project. Am Heart J 2002;143:412–17.[CrossRef][Web of Science][Medline]
  52. Zeger SL, Thomas D, Dominici F, et al. Exposure measurement error in time-series studies of air pollution: concepts and consequences. Environ Health Perspect 2000;108:419–26.[Web of Science][Medline]
  53. Janssen NA, de Hartog JJ, Hoek G, et al. Personal exposure to fine particulate matter in elderly subjects: relation between personal, indoor, and outdoor concentrations. J Air Waste Manag Assoc 2000;50:1133–43.[Web of Science][Medline]
  54. Sarnat JA, Koutrakis P, Suh HH. Assessing the relationship between personal particulate and gaseous exposures of senior citizens living in Baltimore, MD. J Air Waste Manag Assoc 2000;50:1184–98.[Web of Science][Medline]
  55. Opasich C, Rapezzi C, Lucci D, et al. Precipitating factors and decision-making processes of short-term worsening heart failure despite "optimal" treatment (from the IN-CHF Registry). Am J Cardiol 2001;88:382–7.[CrossRef][Web of Science][Medline]
  56. Mittleman MA, Verrier RL. Air pollution: small particles, big problems? Epidemiology 2003;14:512–13.[CrossRef][Web of Science][Medline]
  57. Bateson TF, Schwartz J. Who is sensitive to the effects of particulate air pollution on mortality? A case-crossover analysis of effect modifiers. Epidemiology 2004;15:143–9.[CrossRef][Web of Science][Medline]
  58. Schiff GD, Fung S, Speroff T, et al. Decompensated heart failure: symptoms, patterns of onset, and contributing factors. Am J Med 2003;114:625–30.[CrossRef][Web of Science][Medline]
  59. Friedman MM. Older adults' symptoms and their duration before hospitalization for heart failure. Heart Lung 1997;26:169–76.[CrossRef][Web of Science][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?

Related articles in Am. J. Epidemiol.:

Invited Commentary: Ripeness Is All
Joel Schwartz
Am. J. Epidemiol. 2006 164: 434-436. [Extract] [FREE Full Text]  



This article has been cited by other articles:


Home page
 Occup. Environ. Med.Home page
M S Goldberg, N Giannetti, R T Burnett, N E Mayo, M-F Valois, and J M Brophy
A panel study in congestive heart failure to estimate the short-term effects from personal factors and environmental conditions on oxygen saturation and pulse rate
Occup. Environ. Med., October 1, 2008; 65(10): 659 - 666.
[Abstract] [Full Text] [PDF]

Home page
 Eur Respir JHome page
I. Annesi-Maesano, F. Forastiere, N. Kunzli, B. Brunekref, and on behalf of the Environment and Health Committee
Particulate matter, science and EU policy
Eur. Respir. J., March 1, 2007; 29(3): 428 - 431.
[Full Text] [PDF]

Home page
 Am J EpidemiolHome page
J. Schwartz
Invited Commentary: Ripeness Is All
Am. J. Epidemiol., September 1, 2006; 164(5): 434 - 436.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
164/5/421    most recent
kwj206v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (7)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Symons, J. M.
Right arrow Articles by Geyh, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Symons, J. M.
Right arrow Articles by Geyh, A.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?