Longitudinal trends in asthma emergency department visits, pollutant and pollen levels, and weather variables in the Bronx from 2001–2008

Abstract Objective: To evaluate how asthma-related emergency department visits (AREDV), air pollutant levels, pollen counts, and weather variables changed from 2001 to 2008 in the Bronx, NY. Methods: 42,065 daily AREDV values (1 January 2001 to 31 December 2008) were collected using our institution’s Clinical Looking Glass software. Daily values of sulfur dioxide (SO2), carbon monoxide (CO), ozone (O3), nitrogen dioxide (NO2), temperature, and humidity were obtained from the National Climatic Data Center's Bronx station. Daily tree pollen counts were obtained from the Armonk counting station near the Bronx. Median values for each variable were analyzed using the Mann-Whitney test to compare 2001–2004 and 2005–2008 values. Simple linear regression examined associations between AREDV and individual pollutants. Due to seasonal variations of the variables, each season was considered separately. Results: There were significant decreases for AREDV, SO2, CO, and humidity for all seasons, and for NO2 in the spring and winter. Significant increases occurred for O3 in the spring, fall, and winter; for temperature in the summer and winter; and for tree pollen in the spring. Significant positive associations were found between AREDV and SO2, CO, NO2, and humidity, respectively, while significant negative associations were found between AREDV and O3 and temperature, respectively. Conclusions: From 2001 to 2008, significant: a) decreases in AREDV, SO2, CO, and humidity for all seasons, and decreases in NO2 for the spring and winter; and b) increases in O3, temperature, and spring tree pollen were observed. By tracking and anticipating environmental and pollutant changes, efforts can be made to minimize AREDV.


Introduction
Asthma is a common disease that affects 24.6 million children and adults in the USA [1]. The prevalence of asthma and the number of asthma-related emergency department visits (AREDV) in the Bronx, NY significantly surpass the national rates [2]. The prevalence of asthma in the Bronx is 13.0% [3] in comparison to the national rate of 7.7% [4]. The number of AREDV in the Bronx is 231.4 [5] per 10,000 residents, which greatly exceeds the 69.7 [6] visits per 10,000 residents nationally.
Air pollutants increase the severity of asthma and the number of AREDV [7]. A previous study indicated that an increase in pollutants including particulate matter with a diameter of 2.5 lm or less (PM 2.5 ), particulate matter with a diameter of 10 lm or less (PM 10 ), sulfur dioxide (SO 2 ), carbon monoxide (CO), and nitrogen dioxide (NO 2 ), along with a decrease in ozone (O 3 ) were linked to an increase in AREDV [8]. While the study observed how these pollutants and AREDV were associated, it did not evaluate how longterm pollutant and AREDV values were associated and changed over time. Other studies also observed changes in pollutants including SO 2 , O 3 , NO 2 , and pollen and assessed how they were associated with AREDV rates over a relatively short period of time (one year for the first study, and four years for the second study) [9,10].
Daily pollutant values were previously collected in the Bronx, NY from the years of 2001-2008 (i.e. 1 January 2001-31 December 2008). These pollutants include PM 2.5 , PM 10 , SO 2 , CO, O 3 , NO 2 , and tree pollen. Particle pollutants (PM 2.5 and PM 10 ) are a mixture of particles of varying diameter found in the air that are known to cause decreased lung function, asthma exacerbations, along with onset and progression of other lung diseases [11]. SO 2 is a colorless pollutant emitted from fossil fuel consumption and industrial processes that causes respiratory irritation, respiratory dysfunction, and bronchitis [11]. CO is a colorless gas produced from fossil fuels, such as the burning of coal and wood, which has an affinity to hemoglobin that is much greater than that of oxygen. This allows CO to displace oxygen in red blood cells and no longer allow oxygen to be carried to the organs throughout the body, which can result in respiratory distress [11]. O 3 is a colorless gas found in large amounts in the earth's atmosphere. Ground-level O 3 is formed from natural sources and human activity and is believed to be a plausible cause of increased respiratory diseases, such as asthma [11]. NO 2 is a common pollutant produced from road traffic, as they are emitted from motor-engines in the exhaust. It is known to have an effect of lung damage, lung inflammation, coughing, wheezing, and respiratory infections [11,12]. Tree pollen is a common allergen found in the air that has been associated with severe exacerbations of asthma [13]. Data analysis in this current study was used to understand how the pollutant values changed throughout the years. Any associations found when comparing the change in pollutant values to the change in daily adult and pediatric AREDV values collected during the same time period were then evaluated. Using this information, any association between changes in each environmental factor and changes in AREDV over time can be better understood. Our goal was to evaluate daily pollutant values within the Bronx, NY from 2001-2008 and assess how changes in those values are associated with changes in AREDV.

Participants and background
Participants in this retrospective cross-sectional study included both adult and pediatric patients (i.e. individuals under 18 years of age) of all genders requiring AREDV, leading to a total of 42,065 adult and pediatric AREDV being collected. The aggregate number of daily AREDV at two major Bronx hospitals (Montefiore Medical Center's Moses and Weiler divisions) were obtained and analyzed from 1 January 2001 to 31 December 2008. Daily values of SO 2 , CO, O 3 , NO 2 , temperature, humidity, and tree pollen were collected and analyzed during the same time period. Due to missing data on PM 2.5 and PM 10 , these pollutant values were not included in our analysis. Only the daily number of AREDV and no personally identifiable information were collected and used for the statistical analysis. The study, due to our collection and analysis of de-identified data, was determined to be exempt by the Einstein-Montefiore Institutional Review Board.

Asthma-related emergency department visits
The number of daily AREDV from 2001-2008 were identified using our institution's Clinical Looking Glass (CLG V R ) software. CLG V R is a user-friendly interactive software application developed to evaluate health care quality, effectiveness, and efficiency. The system integrates clinical and administrative datasets allowing non-statisticians to produce epidemiologically cogent reports in order to globally assess care quality. Pediatric and adult patients were identified with a primary diagnosis of asthma (International Classification of Diseases, Ninth Revision, ICD-9 493.0) upon discharge from the emergency department (ED) to yield the aggregate daily AREDV values. As a result of the variations of AREDV between the seasons, the AREDV numbers were considered separately for the four seasons. The winter consisted of dates ranging from 21st December to 19th March. The spring ranged from 20th March to 20th June. The summer spanned from 21st June to 21st September and the fall included dates from 22nd September to 20th December. The AREDV numbers were grouped into two four-year time periods (2001-2004 and 2005-2008) in order to observe how the data varied across the eight-year period.

Pollutant, weather, and pollen values
Daily values of SO 2 (in ppb), CO (in ppm), O 3 (in ppm), NO 2 (in ppb), temperature (in F) and humidity (in %) were obtained from the National Climatic Data Center's (NCDC) Bronx Station from 1st January 2001 to 31st December 2008. Daily tree pollen counts (in grains/m 3 of air) were obtained from the Armonk counting station near the Bronx. As a result of seasonal variations in pollutant and weather-related values, data were analyzed separately for each of the four seasons.

Statistical analyses
Individual, seasonal graphs were created for SO 2 Figures 1 and 2). The winter, spring, summer, and fall seasons all had significant statistical     2005-2008. In the summer, the median value decreased, although the decrease was not statistically significant. The fall median value remained the same and was also not statistically significant ( Table 2).

Spring tree pollen: 2001-2004 vs. 2005-2008
Spring tree pollen values peaked every late-April to early-May throughout the eight-year time period (Supplemental Figures 5 and 6). Overall, the tree pollen increased from 2001-2004 to 2005-2008. In the spring, the median tree pollen value showed a statistically significant increase (Table 2).

Seasonal temperature: 2001-2004 vs. 2005-2008
In the winter and summer seasons, the median temperature showed a statistically significant increase from 2001-2004 to 2005-2008. The spring median temperature remained the same and the fall median temperature increased, although there was no statistical significance (Table 3).  (Table 3).

Simple linear regression analysis
Both SO 2 and CO had a significant positive association with AREDV in all four seasons, however the strongest association was found in the spring season for SO 2 and in the winter season for CO. O 3 was negatively associated with AREDV for the spring, fall, and winter seasons and the strongest association occurred during the winter. However, O 3 was positively associated with AREDV in the summer season. This positive association between O 3 and AREDV was comparatively less strong than the negative associations of the two variables for the other seasons. While NO 2 had a significant positive association with AREDV in all seasons except for spring, tree pollen was positively associated with AREDV only in the spring season. Finally, AREDV was found to be negatively associated with temperature in the spring and fall seasons and had a significant positive association with humidity in the winter season (Table 4).

Discussion
This study investigated the association between changes in daily pollutant values, temperature, humidity, and daily AREDV values from 2001-2008. Our previous research found a correlation between increased spring tree pollen counts and increased AREDV [14]. In this study, tree pollen levels peaked in the spring as well as an overall significant increase in spring tree pollen levels from  [8]. Decreases in SO 2 , CO and NO 2 values, and an increase in O 3 values, which associated with a decrease in AREDV values were found in this study as well. Prunicki et al. reported that exposure to high CO and NO 2 air pollutant levels resulted in DNA methylation of Foxp3 promoter regions within genes, which is associated to an increased prevalence of asthma [16]. A decrease in CO and NO 2 values was also found to be associated with a decrease in AREDV values in this study. Mazenq et al. reported that exposure to high levels of NO 2 have significant correlations to AREDV [17]. Evans et al. reported that there is an inverse association between AREDV and O 3 levels in the environment [18]. Both of these studies support the increase in AREDV being associated to increased levels of NO 2 and decreased levels of O 3 . However, some studies have reported a significant positive association between AREDV/asthma exacerbation and O 3 [10,19]. While an increase in O 3 did not have a direct association to an increase in AREDV in our study, this does not mean that O 3 is a pollutant that should be inhaled. These results only suggest that O 3 and AREDV levels are not associated; however, O 3 has been reported to have associations with other diseases [20,21].
Our study has several limitations. While daily values collected for each of the variables were used in the study, there were some variables that had missing or incomplete data, such as for PM 2.5 and PM 10 . Thus, the effect of these pollutants on AREDV could not be accurately assessed and reported. Another limitation is the small number of stations through which our data were obtained (an NCDC-Bronx station for weather and pollutant variables and an Armonk station for pollen values), which would not give as accurate of a representation of the pollutant levels in the Bronx, NY as more stations would. The number of hospitals (two) that were used to collect our AREDV data is another limitation. A larger sample size that spanned a greater radius of location would provide a better, more balanced set of results. In addition, the AREDV data did not account for other triggers of asthma requiring ED visits, such as home environmental triggers (mold, rodents) and viral respiratory infections. Thus, there is no way to entirely know that the change in pollutant levels are the only factors associated with the change in AREDV over time.
Regarding the data collected, individual level data were not able to be obtained, which consequently did not allow for multivariate analysis or analysis by factors such as age, gender, disease severity or treatment. Instead, simple linear regression analysis was used to assess associations between pollutant values and AREDV. Future studies including individual level data, such as age and gender, could examine possible effects of these variables on study outcomes. Lastly, this study described data collected retrospectively over the eight-year period in the Bronx, NY. Future studies collecting prospective data in other regions of the USA are needed, as these eight years potentially may not reflect the present-day pollutant levels and environmental factors found in the USA.

Conclusion
This study revealed an association between decreased AREDV and decreased NO 2 , CO, and SO 2 values in the Bronx. There was also a statistically significant association between decreased AREDV and increased O 3 values, which shows that although O 3 is a pollutant that should not be inhaled, it does not have a positive association with AREDV. There may be a link between the decrease in humidity and in AREDV that was seen across all seasons throughout the eightyear time period; however, simple linear regression confirmed a significant positive association in the winter season only. By tracking and predicting future environmental and air pollutant changes, efforts can be made to notify the public, adjust clinical practice accordingly, and reduce asthma-related ED visits.  Significance was assessed if p values were 0.05.

Disclosure statement
The simple linear regression model considers AREDV as the dependent variable and only one pollutant/environmental factor as the independent variable.