The association between exposure to air pollution and acute cardiovascular (CV) events is well documented; however, limited data are available evaluating the public health safety of various “doses” of particular matter (PM) below currently accepted safety thresholds. We explored the cross-sectional association between PM with aerodynamic diameter <10 μm (PM10) and daily CV hospitalizations in Brescia, Italy, using Poisson regression models adjusted for age, gender, and meteorologic indices. Average daily exposure to PM10 obtained from arithmetic means of air pollution data were captured by 4 selected monitoring stations. PM10 data were expressed as daily means (lag 0-day) or 3-day moving averages (lag 3-day) and categorized according to the European Union daily limit value of 50 μg/m 3 . From September 2004 to September 2007, data from 6,000 acute CV admissions to a tertiary referral center were collected. An increase of 1 μg/m 3 PM10 at lag 0-day was independently associated with higher rates of acute hospitalizations for composite CV-related events (relative risk [RR] 1.004, 95% confidence interval [CI] 1.002 to 1.006), acute heart failure (RR 1.004, 95% CI 1.001 to 1.008), acute coronary syndromes (RR 1.002, 95% CI 0.999 to 1.005), malignant ventricular arrhythmias (RR 1.004, 95% CI 0.999 to 1.010), and atrial fibrillation (RR 1.008, 95% CI 1.003 to 1.012). Similar results were obtained using PM10 lag 3-day data. The excess PM10 CV hospitalization risk (by lag 0-day and lag 3-day) did not vary significantly above and below the 50 μg/m 3 safety threshold or by age and gender. In conclusion, increased levels of PM10, even below the current limits set by the European Union, were associated with excess risk for admissions for acute CV events.
Over the last several decades, rapid urbanization, motorization, and industrialization have led to a marked rise in air pollution. Despite ongoing efforts by local and global environmental and social agencies to improve air quality, the negative health impacts of air pollution remain major public health concerns. Major mechanisms driving inhalation-mediated cardiovascular (CV) toxicity include activation of proinflammatory pathways and generation of reactive oxygen species. Based on early clinical and experimental data, these adverse biologic effects may directly influence heart rate, heart rate variability, blood pressure, vascular tone, procoagulability, and progression of atherosclerosis.
Air pollution consists of complex mixtures of organic and inorganic toxic compounds, both in gaseous and particulate phases. Particulate matter with aerodynamic diameter <10 μm (PM10) has been associated with CV and respiratory morbidity and mortality and, thus, represents a validated metric of air pollution. Recognizing this substantial CV risk, the European Union (EU) has established air quality standards and set a legal allowable limit for various pollutants in ambient air, including PM10 (<50 μg/m 3 averaged over a 24-hour period [directive 99/30/CE]). However, limited data are available to determine if these thresholds and regulations provide an adequate margin of safety for the general population and for more susceptible subgroups. The aims of the present study were to (1) describe the distribution and variability of PM10 levels measured from a highly industrialized town in Northern Italy over a 3-year period, (2) examine the risk of exposure to PM10 on hospitalizations for acute CV events to a single tertiary referral center, and (3) determine if this CV hospitalization risk varies above and below the established EU safety threshold or by age and gender.
Methods
From September 1, 2004, to September 30, 2007, this cross-sectional observational study assessed the association between PM10 levels and hospitalizations for acute CV events in a city with a high degree of industrialization and urbanization (∼332 inhabitants/km 2 ). The average daily concentrations of PM10 were retrieved from the official Web site of the regional environmental agency (Agenzia Regionale per la Prevenzione e Protezione Ambientale) of the Lombardy region of Northern Italy ( www.arpalombardia.it ), which provides PM10 concentrations registered in urban and suburban areas of the district of Brescia (covering ∼1.2 million inhabitants). The daily PM10 average concentrations of this area were calculated as arithmetic means of concentrations measured by 4 monitoring stations (site A: Broletto; site B: Rezzato; site C: Sarezzo; and site D: Odolo) that, according to Italian law, were not close to local sources of air pollution, such as factories, industries, and high-traffic roads.
Using the Hospital Discharge Database, which includes date of admission and discharge from the hospital, the primary discharge diagnosis, classified according to the International Classification of Diseases, Ninth Revision (ICD-9 ), and the diagnosis-related group code, we identified patients who experienced hospitalizations for CV disease during the study period. Consistent with previous data, hospitalizations for acute heart failure (AHF; ICD-9 428), acute coronary syndromes (ACS; ICD-9 410 to 413), malignant ventricular arrhythmias and/or implantable cardioverter-defibrillator requirement (MVA; ICD-9 427.1, 427.41, and 427.42), and atrial fibrillation (AF; ICD-9 427.31 and 427.32) were extracted.
The cross-sectional association between daily average concentrations of PM10 and hospitalization rates for acute CV events was analyzed by a generalized linear model. We applied the Poisson regression with the log-link function because daily counts of hospitalizations were assumed to follow a Poisson distribution. Concentrations of PM10 were analyzed as a daily 24-hour average (lag 0-day) and as a 3-day moving average (lag 3-day). In each model, lag 0-day pollution effects were evaluated using the arithmetic means of PM10 values obtained on the same admission day, whereas lag 3-day effects were investigated using the arithmetic means of values obtained on the same day and from the 3 days preceding the admission date. This time-series approach is expected to attenuate day-to-day variation of the exposure variable and minimize the influence of changes in individual-level CV-risk profiles of admitted patients. To evaluate the influence of the EU daily limit, lag 0-day PM10 values were also assessed as a dichotomous variable above and below 50 μg/m 3 .
Seasonal variation and specific meteorologic factors are known to influence CV risk. As such, all models were adjusted for age, gender, previous CV hospitalizations, and meteoclimatic variables (mean daily air temperature [°C], relative humidity [%], atmospheric pressure [mbar], and wind speed [km/h]). Age was expressed as a dichotomous variable (<65 or ≥65 years). Contemporaneous data regarding meteorologic variables were obtained from the National Meteorological Institute of the Air Force located in Ghedi (Brescia, Italy) ( www.meteoam.it ). Interaction analyses were performed to determine if the CV hospitalization risk associated with PM10 was modified by the EU safety limit or by age and gender.
Effect sizes were expressed as relative risks (RR) and 95% confidence intervals, determined by Poisson regression models. Means and SD of admissions were calculated according to the Poisson distribution, that is, SD were obtained as square roots of means. Covariates in the final models were age, gender, meteorologic indices (temperature, relative humidity, wind speed, rain, and atmospheric pressure), and the PM10 × age and PM10 × gender interaction terms. Normality of the PM10 distribution was tested based on Kolmogorov-Smirnov testing; p <0.05 was considered statistically significant. All statistical analyses were performed using IBM PASW 19 (IBM Corporation, Armonk, New York).
Results
From September 1, 2004, to September 30, 2007, the distribution of PM10 values was log-normal. The geometric mean of lag 0-day PM10 values was 40.2 (1.8) μg/m 3 (range fifth percentile 15 μg/m 3 and ninety-fifth percentile 108.5 μg/m 3 ). The corresponding geometric mean of lag 3-day PM10 values was 41.3 (1.7) μg/m 3 (range fifth percentile 17.3 μg/m 3 and ninety-fifth percentile 102.1 μg/m 3 ). Of the total 1,125 days surveyed, 413 (36.7%) included PM10 values (based on lag 0-day measurements) above the EU safety threshold of 50 μg/m 3 .
During the study time-frame, there were a total of 6,000 urgent CV admissions: 23% with a primary diagnosis of AHF, 54% with ACS, 9% with MVA, and 14% with AF ( Table 1 ). Older patients and men were more likely to be admitted for all CV events, except MVA. In terms of meteorologic parameters, temperature was negatively associated with all causes of CV admissions, with the exception of AF. Atmospheric pressure was negatively associated with AHF and MVA, whereas rain was positively associated with AF ( Table 2 ).
| Diagnoses | Men n=4,201 | Women n=1,799 | Total n=6,000 |
|---|---|---|---|
| Acute heart failure | 902 (21%) | 484 (27%) | 1,386 (23%) |
| Acute coronary syndrome | 2,423 (58%) | 830 (46%) | 3,253 (54%) |
| Malignant ventricular arrhythmia | 316 (8%) | 215 (12%) | 531 (9%) |
| Atrial fibrillation | 560 (13%) | 270 (15%) | 830 (14%) |
| Variables | Acute heart failure | Acute coronary syndrome | Malignant ventricular arrhythmia | Atrial fibrillation | ||||
|---|---|---|---|---|---|---|---|---|
| RR | 95% CI | RR | 95% CI | RR | 95% CI | RR | 95% CI | |
| Male | 1.864 ∗∗∗ | 1.669-2.081 | 2.919 ∗∗∗ | 2.698-3.159 | 1.470 ∗∗∗ | 1.236-1.748 | 2.074 ∗∗∗ | 1.794-2.398 |
| Age <65 years | 0.535 ∗∗∗ | 0.479-0.597 | 0.860 ∗∗∗ | 0.803-0.921 | 1.004 | 0.847-1.190 | 0.774 ∗∗∗ | 0.674-0.887 |
| Temperature (°C) | 0.988 ∗∗ | 0.980-0.996 | 0.990 ∗∗∗ | 0.985-0.995 | 0.967 ∗∗∗ | 0.954-0.980 | 0.999 | 0.989-1.010 |
| Relative humidity (%) | 0.998 | 0.994-1.003 | 0.999 | 0.996-1.001 | 0.995 | 0.988-1.001 | 0.995 | 0.990-1.001 |
| Wind speed (km/h) | 1.029 | 0.994-1.066 | 0.982 | 0.959-1.005 | 1.003 | 0.946-1.063 | 1.016 | 0.971-1.063 |
| Rain (mm) | 0.996 | 0.972-1.022 | 1.005 | 0.989-1.021 | 0.987 | 0.943-1.032 | 1.036 ∗ | 1.007-1.065 |
| Atmospheric pressure (mbar) | 0.992 ∗ | 0.984-1.00 | 0.999 | 0.993-1.004 | 0.988 ∗ | 0.976-1.000 | 0.999 | 0.989-1.009 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
