Environ Sci Pollut Res (2015) 22:14651–14662 DOI 10.1007/s11356-015-5188-x

REVIEW ARTICLE

Short-term exposure to particulate air pollution and risk of myocardial infarction: a systematic review and meta-analysis Chunmiao Luo 1 & Xiaoxia Zhu 2 & Cijiang Yao 2 & Lijuan Hou 2 & Jian Zhang 2 & Jiyu Cao 3 & Ailing Wang 4

Received: 27 February 2015 / Accepted: 10 August 2015 / Published online: 23 August 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract A growing number of studies have associated short-term exposure to ambient particulate matter air pollution (PM) and risk of specific cardiovascular events, just as myocardial infarction (MI). However, the results of the recent studies were inconsistent; therefore, a systematic review and meta-analysis was performed. To synthetically quantify the association between short-term exposure to PM and risk of MI, a meta-analysis was conducted to combine the estimates of effect for a relationship between short-term exposure to PM10, PM2.5 (particulate matter≤10 μm, 2.5 μm in diameter) and risk of MI. Electronic database searches for all relevant published studies were updated in January 2015. And, a random-effects model was performed to estimate pooled relative risk (RR) and 95 % confidence intervals (95 % CI). Thirty-one published observational epidemiological studies

were identified. Risk of MI was significantly associated with per 10 μg/m3 increment in PM10 (OR=1.005; 95 % CI 1.001– 1.008) and PM2.5 (OR=1.022; 95 % CI 1.015–1.030). The risk of PM2.5 exposure was relatively greater than PM10. In the subgroup analysis by study design, location, quality score, and lag exposure, the results were basically consistent with the former overall results in PM2.5 but slightly changed in PM10. Short-term exposure to particulate matter (PM2.5, PM10) was a risk factor for MI, and the results further confirmed the discovery in the previous meta-analysis. Keywords Ambient particulate matter . Meta-analysis . Myocardial infarction . Short term

Introduction Responsible editor: Philippe Garrigues Chunmiao Luo and Xiaoxia Zhu contributed equally to this work. Electronic supplementary material The online version of this article (doi:10.1007/s11356-015-5188-x) contains supplementary material, which is available to authorized users. * Ailing Wang [email protected] 1

Department of Cardiology Medical, The Second People’s Hospital of Hefei, Heping Road, Hefei, Anhui, China

2

Department of Occupational and Environmental, School of Public Health, Anhui Medical University, Meishan Road, Hefei, Anhui, China

3

The Teaching Center for Preventive Medicine, School of Public Health, Anhui Medical University, Meishan Road, Hefei, Anhui, China

4

Department of Cardiology Medical, the First Affiliated Hospital of Anhui Medical University, Jixi Road, Hefei, Anhui, China

Myocardial infarction (MI) is an acute, severe, and specific cardiovascular events and is the leading cause of mortality among both women and men in industrialized countries (Cendon et al. 2006; Bhaskaran et al. 2009). Identifying relevant risk factors for MI and then taking effective measures to reduce the effect is of significant public health problem. It is well established that risk factors for MI include hypertension, diabetes, obesity, tobacco use, high cholesterol levels, physical inactivity, and psychosocial factors (Castelli 1984; Yusuf et al. 2004). And, a large body of studies has associated shortterm exposure to daily average levels of air pollution and risk of MI (Hoffmann et al. 2009; Turin et al. 2012; Jalaludin and Cowie 2014; Talbott et al. 2014; Wichmann et al. 2014; Xie et al. 2014). In particular, airborne particulate matter (PM) not only is an extremely complex air pollutant but also is a large harmful risk factors to human health. Whereas over the past decade, a growing number of studies have been focusing on the hazard of coarse ambient particulate matter, aerodynamic

14652

diameter ≤10 μm (PM10) and fine ambient particulate matter, aerodynamic diameter ≤2.5 μm (PM2.5) (Barnett et al. 2006; Hodas et al. 2013; Bard et al. 2014). Especially, the composition of PM2.5 always changes for emitting from a variety of sources and varies temporally and spatially (Heo et al. 2014; Zhu et al. 2015). Furthermore, on account of the large specific surface area, its small size and long residence time in air, more likely adsorbing various harmful substances in the air, PM2.5 can enter the depths of human respiratory tract and participate in the blood circulation; therefore, it tends to be more harmful to human than other air pollutants (Kreyling et al. 2002; Billet et al. 2007; Zhu et al. 2015). An elevated risk of inflammation, oxidative stress, increased blood coagulation, cardiac arrhythmia, changed vascular function, or others could be the potential mechanistic pathways by which exposure to ambient particulate matter air pollution may result in onset of MI (Mills et al. 2009; Brook et al. 2010; Mustafic et al. 2012). It is plausible that short-term exposure to high levels of PM could similarly elevate the risk of MI. Even though a growing number of studies have demonstrated the association between transient elevations in PM10, PM2.5, and risk of MI (Hoffmann et al. 2009; Turin et al. 2012; Jalaludin and Cowie 2014; Talbott et al. 2014; Wichmann et al. 2014; Xie et al. 2014), the results of many time-series studies and case-crossover studies in recent years had conflicting results compared with a recent meta-analysis (Mustafic et al. 2012). To our knowledge, the systematic review and meta-analysis have been published in February 2012, which only have searched relevant literatures up to 2011, comprehensively clarifying the relationship between main air pollution (SO2, NO2, NO, O3, PM10, PM2.5) and risk of MI, and observing an evidence of publication bias in PM2.5 exposure, but not PM10. While in recent years, PM has gained widely attention, and its hazard has also been being researched extensively. And, there are increasing studies on the relationship between short-term exposure to daily average levels of PM10, PM2.5, and risk of MI. In this study, we perform summary estimates of effect on the associations between shortterm exposure to daily average levels of PM10, PM2.5, and risk of MI, including 31 epidemiological studies published throughout January 2015, additional quantifying heterogeneity, conducting subgroup analysis, and evaluating publication bias across studies.

Methods Literature search We searched the electronic databases of PubMed, Web of Science, Embase, and Google Scholar in English for all published studies and evaluated the effect of short-term exposure to PM10 and PM2.5 on the risk of MI, using combinations of

Environ Sci Pollut Res (2015) 22:14651–14662

the following terms: Bambient particulate matter air pollution^ or Bparticle,^ BPM10,^ BPM2.5,^ and Bmyocardial infarction^ or BMI,^ Bacute coronary syndrome.^ No date restrictions were performed to the search engines. All original studies only studied on humans and published in English were included. We also searched additional studies from the reference lists of these identified articles, previous relevant published reviews, and meta-analysis to identify studies not included by the initial search. The selection of original studies was based on the inspection of all the titles and/or abstracts among all the potentially eligible articles. Then, the selection of potentially eligible articles was read entirely to determine that they were suitable for the inclusion and exclusion criteria in the review and meta-analysis.

Quality score assessment As described in a previous review and meta-analysis (Mustafic et al. 2012), reporting quality score assessment for all eligible articles was necessary for time-series studies and case-crossover studies. However, no validated scales are available to evaluate methodological quality at present. Therefore, we adapted a quality scale based on the previous review and meta-analysis and from validated scales of other articles. Two reviewers also evaluated the quality of studies independently on three components, using as follows: (1) ascertainment of MI occurrence (0 to 1 point). A score of 1 was given if the diagnosis of MI was coded in terms of the International Classification of Diseases or based on angiographic criteria, or according to the triad of clinical, laboratory, and electrocardiographic criteria (Mustafic et al. 2012); 0 was given if there was no description of diagnosis or the diagnosis of MI was ascertained by patient history or other criteria. (2) The quality of particulate air pollutant measurements (0 to 1 point). If measurements were performed at least daily and less than 25 % missing data, 1 was given; if measurements were not performed at least daily or more than 25 % missing data (Mustafic et al. 2012) or there was no description of particulate measurements, 0 was given. (3) Adjustment for confounders (0 to 1 point). Due to differences of research methods between time-series study and case-crossover study, the methods of adjustment for confounders were also different. One point was given if adjustment for covariates was performed for several important covariates together, including long-term trends, seasonality, temperature, and humidity/pressure/day of week for time-series studies; for case-crossover studies which control invariant and slowly changing confounders by the design itself, therefore, if an adjustment was made for temperature and humidity/pressure/day of week together, 1 point was given. Without adjusting the above several important covariates together, 0 point was given. If a study got a maximum score in each components, it would be considered

Environ Sci Pollut Res (2015) 22:14651–14662

to be of good quality. And, if one of the three components got the minimum score (0 point), it would be of low quality. Study selection (inclusion and exclusion criteria) Any time-series analysis or case-crossover epidemiologic studies were included in the meta-analysis if they investigated the association of short-term exposure to PM10 and PM2.5 with the risk of MI. Short-term was defined as a few hours to a few days (0–7 days) before the event day. And, if the independent observational epidemiological studies showed a risk estimates of MI (relative risk (RR) and 95 % confidence intervals (95 % CI)) with a unit or IQR (inter-quartile range) change transiently in PM10 and PM2.5 mass concentration, they would be extracted. Studies evaluating the association of long-term exposure to PM with risk of MI were excluded. If a study did not show enough quantitative data, we would contact the author by sending an email, and the study would be excluded without getting an answer. To avoid involving the same or overlapping dataset, when the same patient population in a region was included in several publications, only the study with the largest number of observations and/or the longest study period was included in the meta-analysis (Zhu et al. 2015). For example, results from Zanobetti and Schwartz (2006) were chosen over Peters et al. (2001) in the same region, for the former reporting a longer study period. What’ s more, Zanobetti and Schwartz (2005) that reported the results in 21 US cities from 1985 to 1999 was chosen over Braga et al. (2001) for a larger number of observational population. Data extraction In accordance with the inclusion and exclusion criteria in the study selection, two investigators initially searched and extracted date from all eligible studies independently to ensure the accuracy of the information. For each study, information was extracted as follows: the author, year of publication, country, study period, study design, pollution exposure, average exposure levels, case of population, pollutant models, effect size, and adjustment of other factors and score quality. If any discordance between the two authors existed, a third author would be sought an opinion for obtaining a consensus. Statistical analysis Risk estimates of each included study were converted to RR and their 95 % CI in risk of MI with a short-term common exposure per 10 μg/m3 increase in PM10 and PM2.5. When single and multipollutant models existed in one included study, we selected results from single pollutant. And, when risk estimates were shown for several different lag patterns in the same study, we chose the most frequently used pattern

14653

among the selected studies as the overall analysis. The random-effects model and the fixed-effects model were performed to obtain the combined RR estimate of each included study (DerSimonian and Laird 1986). The statistical heterogeneity among studies was evaluated by using the chi-squarebased Q statistic test (Cochran 1954), a P value 50 % demonstrated a statistically significant heterogeneity, and 25 % or less, 25 to 75 %, and 75 % or more are respectively used to represent low, moderate, and high degree heterogeneity (Mustafic et al. 2012). In this study, heterogeneous of study design, location, and score quality subgroup were also analyzed. We removed a single study included in the meta-analysis each time to reflect the influence of the individual dataset to the pooled RR (Tobias 1999). Publication bias was assessed in a Begg’s test; then, the degree of asymmetry was tested by Egger’s test (Che et al. 2014). If publication bias was proved to be existed, we would additionally apply a trim and fill method (Duval and Tweedie 2000) to make an adjustment for publication bias. We also performed four subgroup analysis including study design (time-series study, case-crossover study), geographic locations (Asia, North America, Europe, Australia), score quality (low quality, high quality), and lag periods (lag0, lag1, lag2, lag3, lag4, lag5, lag6, and lag0-1) to explore the robustness of the results. All of the statistical analyses were performed using the software Stata version 11.0 (StataCorp LP, College Station, TX, USA), and a P value

Short-term exposure to particulate air pollution and risk of myocardial infarction: a systematic review and meta-analysis.

A growing number of studies have associated short-term exposure to ambient particulate matter air pollution (PM) and risk of specific cardiovascular e...
1KB Sizes 0 Downloads 30 Views