Particulate matter exposure predicts residence in high-risk areas for community acquired pneumonia among hospitalized children

Abstract

Particulate matter exposure is a risk factor for lower respiratory tract infection in children. Here, we investigated the geospatial patterns of community-acquired pneumonia and the impact of PM_2.5 (particulate matter with an aerodynamic diameter ≤2.5 µm) on geospatial variability of pneumonia in children. We performed a retrospective analysis of prospectively collected population-based surveillance study data of community-acquired pneumonia hospitalizations among children <18 years residing in the Memphis metropolitan area, who were enrolled in the Centers for Disease Control and Prevention sponsored Etiology of Pneumonia in the Community (EPIC) study from January 2010 to June 2012. The outcome measure, residence in high- and low-risk areas for community-acquired pneumonia, was determined by calculating pneumonia incidence rates and performing cluster analysis to identify areas with higher/lower than expected rates of community-acquired pneumonia for the population at risk. High PM_2.5 was defined as exposure to PM_2.5 concentrations greater than the mean value (>10.75 μg/m³), and low PM_2.5 is defined as exposure to PM_2.5 concentrations less than or equal to the mean value (≤10.75 μg/m³). We also assessed the effects of age, sex, race/ethnicity, history of wheezing, insurance type, tobacco smoke exposure, bacterial etiology, and viral etiology of infection. Of 810 (96.1%) subjects with radiographic community-acquired pneumonia, who resided in the Memphis metropolitan area and had addresses which were successfully geocoded (Supplementary Figure F2), 220 (27.2%) patients were identified to be from high- (n = 126) or low-risk (n = 94) community-acquired pneumonia areas. Community-acquired pneumonia in Memphis metropolitan area had a non-homogenous geospatial pattern. PM_2.5 was associated with residence in high-risk areas for community-acquired pneumonia. In addition, children with private insurance and bacterial, as opposed to viral, etiology of infection had a decreased risk of residence in a high-risk area for community-acquired pneumonia. The results from this paper suggest that environmental exposures as well as social risk factors are associated with childhood pneumonia.

Keywords

Pneumonia pediatrics particulate matter spatial patterns high-risk areas outcomes research

Impact statement

Residence in areas where ambient PM_2.5 concentrations are greater than 10.75 μg/m³, which is less than the Environmental Protection Agency (EPA) allowable cut-point for PM_2.5 (<12 μg/m³), is associated with high-risk for community-acquired pneumonia (CAP) in children.

Introduction

Respiratory conditions are the most frequent reason for non-neonatal hospitalization among US children, and pneumonia is the most common principal diagnosis.¹ Outdoor ambient air pollution is associated with higher rates of lower respiratory tract infections (LRTI) in children.^2–6 Particulate matter (PM) can be generated from a variety of sources including anthropogenic and natural sources; and the contribution of these sources to the total concentration of PM can have significant health impacts.⁷ Combustion sources contribute significantly to most of the PM found in PM_2.5, while dusts comprise a significant portion of PM₁₀. Since PM_2.5 (and PM_0.1 contained within PM_2.5) is most strongly correlated with adverse health effects and children are known to be more vulnerable than adults to the adverse effects of PM_2.5, our data are focused on this category of PM. Specifically, particulate matter less than or equal to 2.5 micrometers in diameter (PM_2.5) is implicated in causing respiratory disease,^4,8,9 and high PM_2.5 exposure has been linked to increased risk for acute respiratory infection hospitalizations^2,3,6,10,11 and prolonged length of stay for pneumonia.^12,13 Particulate matter exposure is measured using monitoring data from on-road mobile emission sources, major industrial facilities emission sources, and a high-resolution surface 1 km by 1 km satellite-derived PM_2.5 data. High-resolution 1 km by 1 km PM_2.5 is emerging as a more sensitive measure for assessing PM_2.5 effects on several health outcomes.^14–18 Biologic and clinical risk factors for pediatric pneumonia have been well studied;^19–22 however, few studies have investigated how geospatial patterns affect community-acquired pneumonia (CAP) risk in children^23–28 and this has not directly been examined in U.S. children. Geographical Information Systems (GIS) provide the opportunity to examine associations between clinical factors, environmental factors, and spatial distribution of disease.^25,29–34 An earlier study demonstrated associations between CAP hotspots hospitalizations for pneumonia and asthma in children under 10 years of age and exposure to air pollution over a moderate resolution grid of cells 30 km × 30 km.¹³ However, our study used a much higher spatial resolution of PM_2.5 grid (1 km × 1 km) to study exposure to air pollution among children under 18 years of age hospitalized with CAP in Memphis.

We used data from the Centers for Disease Control and Prevention (CDC) Etiology of Pneumonia in the Community (EPIC) Study, an active, population-based, surveillance study of pediatric CAP hospitalizations, to assess the geospatial patterns CAP in the Memphis Metropolitan area (MMA) and the impact of PM_2.5 on geospatial variability of pneumonia in children. We hypothesized that geographical heterogeneity exists among pediatric CAP patients in the MMA, and that high PM_2.5 exposure is associated with residence in high-risk CAP areas of the MMA.

Materials and methods

Study population, study design, and settings

The EPIC study was a prospective, population-based, multi-center active surveillance study of the incidence and etiology of CAP hospitalizations in the United States; details have been previously published.³⁵ In brief, between January 2010 and June 2012, children <18 years old were eligible for enrollment in the EPIC study, if they were hospitalized at one of three pediatric study hospitals in three different U.S. cities (Memphis and Nashville, TN; and Salt Lake City, UT); resided within a defined catchment area; and had evidence of clinical and radiographic evidence of pneumonia within 72 h of admission. Children were excluded if they were recently hospitalized (i.e. within the past 90 days), enrolled in the study in the preceding 28 days, severely immunocompromised, had cystic fibrosis or a tracheostomy.³⁵ This analysis is limited to the children enrolled at Le Bonheur Children’s Hospital (LBCH) in Memphis, Tennessee who resided in the MMA (Figure 1). The market share for LBCH for pneumonia for patients living in the study area is estimated to be 86% and is accounted for in data analysis.³⁵ The MMA is made up of eight counties (Arkansas: Crittenden; Mississippi: Desoto, Marshall, Tate, Tunica; Tennessee: Fayette, Shelby, Tipton) comprising a population of approximately 1,320,000 according to the U.S. 2010 Census Bureau. For this analysis, we defined the population at risk as all children <18 years old living in the MMA. The Institutional Review Boards for The University of Tennessee Health Science Center and the CDC approved the study protocol.

Figure 1.

Map showing location of study area including eight major regions of the Memphis Metropolitan Area, EPA monitoring sites, major roads, Superfund sites, and top PM_2.5 stationary emission sites based on tons of PM produced.

Demographic, clinical, and laboratory data were collected by caregiver interview and chart abstraction. Etiology of pneumonia was determined using a combination of samples collected for study purposes and clinical samples collected by treating physicians.³⁵ The methods for data collection, specimen collection, laboratory testing, and definitions for pathogen detection have been previously described.³⁵ For this analysis, patients were placed in two mutually exclusive categories; they had a bacterial etiology if they had detection of any bacteria with or without one or more viruses and a viral etiology if one or more viruses and no bacteria were detected, as defined by the EPIC study.

Residential addresses of CAP cases were geocoded and quality checked using ArcGIS Street Maps (ESRI Inc., Redlands, CA, USA), and any unmatched addresses were resolved using online Google maps (Google Maps/Google Earth, Google Inc. Mountain View, California, USA). Geographical information systems (GIS) mapping and local cluster detection were conducted using both ArcGIS 10.4 and SpaceStat 4.0.21 (ClusterSeer, a BioMedware Inc., Ann Arbor, Michigan). We accounted for inflated CAP rates caused by small numerator or denominator data by adjusting for population at risk using an Empirical Bayesian filter.^36,37 Descriptive statistics, bivariate, and multivariable analyses were conducted using IBM SPSS Software version 24 (IBM SPSS Statistics, Armonk, New York).

Statistical analysis

Using the Memphis EPIC study dataset, we performed spatial analysis to assess if there were geographic clusters of CAP in the MMA, and if high PM_2.5 exposure, along with other risk factors, could predict residence in high-risk versus low-risk CAP areas. All spatial analysis was done either at individual or census tract-level. Count of all children < 18 years of age living in the study region was compiled from the US Census data and served as the denominator (population at risk), while the numerator was the count of pneumonia (PN) cases. The outcome variable, residence in high-risk versus low-risk CAP areas, was determined by “hotspot” analysis using Turnbull’s method ³⁸ which evaluates local spatial clusters using the number of CAP cases, user-defined population threshold, and population size (total population of children <18 years old in each census tract). Incidence calculations, adjusted for market share, were made to assess hotspots and cold spots. The lower and upper limits of the 95% confidence intervals for the incidence were derived using two methods previously described.^39,40 For the first part of presented results, we are using bootstrap methods using 10,000 samples, while for the remaining analyses CI was derived from Chi-square tests and logistic regression. A map showing significant spatial clusters of locations of CAP cases was generated (Figure 2). Identified high- (hotspots) and low-risk (cold spots) areas of CAP incidence were used as the outcome variables. Patients who did not reside in high- or low-risk areas of CAP were not included in the final analysis.

Figure 2.

Map showing, at census tract level, spatial clusters/hotspots and cold spots for community acquired pneumonia cases.

Exposure to PM_2.5 was the primary independent variable of interest. We assessed proximity to PM_2.5 by looking at residence within 500-m radius of major roadways and residence within a 2.5-km radius from individual sources. Residential PM_2.5 exposure was measured using a high spatiotemporal resolution 1 km by 1 km grid of satellite-derived PM_2.5 concentrations (Supplementary Methods) and this was chosen as our primary PM variable of interest (Supplementary Discussion). PM_2.5 data spanning the study period were obtained from a published source.¹⁶ Individual-level residential geocodes for each child (point data) were used to extract a specific PM_2.5 value from the PM_2.5 surface (raster data). Each geocoded address was assigned this PM_2.5 value, which was then compiled as an input variable for the model. The mean PM_2.5 for the MMA during the study period was 10.75 μg/m³ (interquartile range (IQR): 10.30; 10.90; 11.30; 12.90; range 8.20–12.90 μg/m³), which is less than the Environmental Protection Agency (EPA) allowable cut-point for PM_2.5 (<12 μg/m³).⁴¹ We then converted this to a dichotomous variable, higher and lower exposure, based on the mean PM_2.5 concentration of 10.75.

We also assessed the effect of other independent variables including: age (<2 years, 2–4 years, 5–9 years, 10–17 years), sex, race/ethnicity (White, non-Hispanic (NH) Black, Hispanic, and “other” encompassing other races, along with multiracial), history of wheezing, insurance type (private vs. public, other), tobacco smoke exposure (number of household smokers “0,” “1,” or “≥2”), bacterial etiology, and viral etiology of infection. We defined severe CAP as admission to the intensive care unit (ICU), prolonged length of stay (prolonged LOS; hospital stay longer than the median length of stay of 2.8 days), and/or any complication (invasive mechanical ventilation, severe sepsis, parapneumomonic effusion requiring intervention, extracorporeal membrane oxygenation, or continuous renal replacement therapy).

Chi-square tests and logistic regression were used to compute unadjusted and adjusted odds ratios (OR) to describe the association of residential PM_2.5 exposure with residence in a high-risk CAP area. After identification of CAP clusters, a preliminary analysis revealed PM_2.5 as the primary independent variable of interest to include in the model (Supplementary Table 1). Other variables, with P < 0.10, were considered for inclusion in the multivariable model. All comparisons were two-sided and P < 0.05 was considered statistically significant in the final multivariable model.

Results

Characteristics, demographics, and spatial distribution of study sample

The MMA comprises eight counties, with a population of 1,320,000 per the U.S. 2010 Census Bureau (Figure 1). At the Memphis study site, 977 children were enrolled; 842 (86.2%) had radiographic CAP and the specimens necessary to determine the etiology of infection. Of these, 810 (96.2%) resided in the MMA and were successfully geocoded (Supplementary Figure 1). The overall incidence of hospitalized pediatric pneumonia for the MMA was 28.72 cases per 10,000 children (95% confidence interval (CI): 28–30). Rates were higher for Hispanics, 75.2 cases per 10,000 (CI: 63–89) non-Hispanic Blacks, 33 cases per 10,000 (CI: 30–36), and other races, 37 cases per 10,000 (CI: 25–54) compared with White children (4 cases per 10,000 [CI: 3–5]). Most pneumonia cases 83.6% (184/220) were from Shelby County followed by Crittenden County 7.3% (16/220), and Desoto County 5% (11/220). The lowest number of cases was observed in descending order in the following counties: Marshall, Fayette, and Tunica.

Among 810 eligible CAP cases, 220 (27.2%) were identified to be from high- or low-risk CAP areas; 126 (57.3%) were from high-risk CAP areas, and 94 (42.7%) from low-risk CAP areas (Supplementary Figure F1). Figure 2 shows, at census tract level, spatial clusters for CAP in children included in our study. Most (44.1%) children admitted with CAP were <2 years of age (Table 1). The racial distribution of CAP cases was mostly NH-Black (76.3%), with a significantly higher proportion of NH-Black children living in high- compared to low-risk CAP areas (89.9% vs. 58.0%, P <0.01). The majority of children had public insurance (84.1%); and a greater proportion who reside in high-risk CAP areas, had public insurance (95.8% vs. 68.2%, P <0.01) compared to those who reside in low-risk CAP areas. Most of the included CAP cases were exposed to high levels of PM_2.5 (63.8%), and a higher proportion of cases residing in high-risk CAP areas were exposed to high PM_2.5 compared with cases living in low-risk CAP areas (75.6% vs. 47.7%, P <0.01).

Table 1.

Descriptive statistics for children hospitalized with community-acquired pneumonia (CAP) from higher than expected and lower than expected CAP areas (n = 220).^a

Variable	Total (n = 220)	Higher than expected (n = 126)	Lower than expected (n = 94)
Age
<2yrs	97 (44.1)	55 (43.7)	42 (44.7)
2–4yrs	71 (32.3)	43 (34.1)	28 (29.8)
5–9yrs	29 (13.2)	15 (11.9)	14 (14.9)
10–17yrs	23 (10.5)	13 (10.3)	10 (10.6)
Race/ethnicity
White	35 (15.9)	3 (2.4)	32 (34.0)
Hispanic	15 (6.8)	9 (7.1)	6 (6.4)
NH-Black	162 (73.6)	109 (86.5)	53 (56.4)
Other	8 (3.6)	5 (4.0)	3 (3.2)
Sex
Male	121 (55.0)	73 (57.9)	48 (51.1)
Female	99 (45.0)	53 (42.1)	46 (48.9)
History of wheeze
Yes	131 (59.5)	73 (57.9)	58 (61.7)
No	89 (40.5)	53 (42.1)	36 (38.3)
Smokers in home
0	113 (51.4)	37 (54.4)	53 (56.4)
1	65 (29.5)	18 (26.5)	24 (25.5)
≥2	42 (19.1)	13 (19.1)	17 (18.1)
Insurance
1 (public)	181 (82.3)	119 (94.4)	62 (66.0)
2 (private)	34 (15.5)	5 (4.0)	29 (30.9)
3 (other)	5 (2.3)	2 (1.6)	3 (3.2)
Year admitted
2010	74 (33.6)	40 (31.7)	34 (36.2)
2011	82 (37.3)	46 (36.5)	36 (38.3)
2012	64 (29.1)	40 (31.7)	24 (25.5)
Season of hospitalization
Spring	63 (28.6)	40 (31.7)	23 (24.5)
Summer	30 (13.6)	15 (11.9)	15 (16.0)
Fall	64 (29.1)	38 (30.2)	26 (27.7)
Winter	63 (28.6)	33 (26.2)	30 (31.9)
1 km × 1 km surface PM_2.5
Low	81 (36.8)	32 (25.4)	38 (40.4)
High	139 (63.2)	94 (74.6)	56 (59.6)
500-m radius from major roadways
No	37 (16.8)	13 (10.3)	24 (25.5)
Yes	183 (83.2)	113 (89.7)	70 (74.5)
2.5-km radius from individual sources
No	144 (65.5)	62 (49.2)	82 (87.2)
Yes	76 (34.5)	64 (50.8)	12 (12.8)
ICU admission
No	191 (86.8)	109 (86.5)	82 (87.2)
Yes	29 (13.2)	17 (13.5)	12 (12.8)
PLOS
No	118 (53.6)	71 (56.3)	47 (50.0)
Yes	102 (46.4)	55 (43.7)	47 (50.0)
Complications
No	210 (95.5)	121 (96.0)	89 (94.7)
Yes	10 (4.5)	5 (4.0)	5 (5.3)
Bacterial etiology
No	196 (89.1)	115 (91.3)	81 (86.2)
Yes	24 (10.9)	11 (8.7)	13 (13.8)
Viral etiology
No	54 (24.5)	28 (22.2)	26 (27.7)
Yes	166 (75.5)	98 (77.8)	68 (72.3)
Any infection
No	28 (12.7)	15 (11.9)	13 (13.8)
Yes	192 (87.3)	111 (88.1)	81 (86.2)

NH: non-Hispanic; PLOS: prolonged length of stay (i.e. >2.8 days); ICU: intensive care unit stay.

^aData presented as n (%).

Bivariate analyses and multivariate analyses were conducted with a reduced sample size (n = 207) due to excluded cases belonging to categories where cell size was <10 (Supplementary Methods). In the bivariate analysis (Table 2), compared with children with CAP in low-risk CAP areas, children with CAP in high-risk areas were more likely to be NH-Black (OR 21.7; 95% CI 6.33–74.25) and Hispanic (OR 15.5; 95% CI 3.2–74.7). In addition, compared with children with CAP in low-risk areas, children with CAP in high-risk areas were less likely to have private insurance (OR 0.09; 95% CI 0.04–0.26) and were more likely to be exposed to high PM_2.5 (OR 3.4; 95% CI 1.9–6.1).

Table 2.

Bivariate analysis for risk factors associated with residence in a higher than expected vs. lower than expected community acquired pneumonia (CAP) areas (n = 207).^a

Variable	OR (95% CI)	P-value
Age		0.94
<2yrs	Reference
2–4yrs	1.19 (0.63–2.26)	0.59
5–9yrs	0.97 (0.41–2.29)	0.94
10–17yrs	0.95 (0.36–2.50)	0.91
Race/ethnicity^b		<0.01
White	Reference
Hispanic	15.50 (3.22–74.66)	0.001
NH-Black	21.68 (6.33–74.25)	<0.01
Sex
Male	Reference
Female	0.82 (0.47–1.43)	0.31
History of wheeze
No	Reference
Yes	1.14 (0.65–1.99)	0.66
Smokers in home		0.36
0	Reference
1	1.60 (0.84–3.04)	0.16
≥2	1.21 (0.57–2.55)	0.62
Insurance^b^,c
Public	Reference
Private	0.09 (0.04–0.26)	<0.01
Season		0.72
Spring	Reference
Summer	0.64 (0.26–1.57)	0.33
Fall	0.83 (0.40–1.74)	0.63
Winter	0.71 (0.34–1.47)	0.35
1 km × 1 km surface PM_2.5^b^,c
Low	Reference
High	3.40 (1.88–6.14)	<0.001
ICU
No	Reference
Yes	0.91 (0.40–2.06)	0.83
PLOS
No	Reference
Yes	0.84 (0.48–1.46)	0.54
Complications
No	Reference
Yes	0.58 (0.15–2.22)	0.42
Bacterial etiology
No	Reference
Yes	0.60 (0.052–1.40)	0.23
Viral etiology
No	Reference
Yes	1.41 (0.74–2.68)	0.30

NH: non-Hispanic; PLOS: prolonged length of stay (i.e., >2.8 days)

^an = 207 (excludes 13 cases with small sample size).

^bStatistically significant P <0.05.

^cIncluded in final model (P <0.1).

On multivariate analysis, we have shown two models (Tables 3 and 4). The first model includes PM_2.5 exposure as the primary explanatory variable (Table 3). This model showed that compared to children with CAP in low-risk CAP areas, children with CAP in high-risk areas were more likely to have high PM_2.5 exposure (aOR 2.47; 95% CI 1.28–4.76), and less likely to have private insurance (aOR 0.12, CI 0.04–0.34). The second model includes race and showed that compared with children with CAP in low-risk CAP areas, children with CAP in high-risk areas were more likely to be NH-Black (aOR 9.821; 95% CI 2.62–36.84) or Hispanic (aOR 6.43; 95% CI 1.20–34.60) versus White children, and were less likely to have private insurance (aOR 0.178; 95% CI 0.059–0.537). Children residing in high-risk CAP areas were also more likely to have high PM_2.5 exposure (aOR 1.66; 95% CI 0.81–3.38), though not statistically significant.

Table 3.

Multivariable model results for residence in higher than expected versus lower than expected CAP areas with PM_2.5 as the primary independent variable (reduced model).

Variable	OR (95% CI)	P-value
1 km × 1 km surface PM_2.5
Low	Reference
High	2.47 (1.31–4.66)	0.005
Insurance
Public	Reference
Private	0.12 (0.04–0.35)	<0.001

Table 4.

Multivariable model results for residence in higher than expected versus lower than expected CAP areas (full model).

Variable	OR (95% CI)	P-value
Insurance
1 (public)	Reference
2 (private)	0.178 (0.059–0.537)	0.002
Bacterial etiology (no vs. yes)	0.183 (0.053–0.634)	0.007
Race/ethnicity
White	Reference	0.003^$
Hispanic	6.434 (1.197–34.595)	0.030
NH-Black	9.821 (2.618–36.838)	0.001
1 km × 1 km surface PM_2.5(0 vs 1)	1.655 (0.810–3.379)	0.167

^$Test for heterogeneity.

Discussion

In our prospective study of CAP in a metropolitan, racially diverse city with varying PM_2.5 exposure, we found that PM_2.5 levels are associated with geospatial patterns of CAP incidence and residence in CAP hotspot areas for children with CAP; however, race is a more significant factor associated with living in high-risk CAP areas. Insurance type is also associated with residence in high-risk areas for CAP in the MMA. Geospatial patterns of CAP incidence were concentrated among children living in north, south, and downtown in the City of Memphis and Shelby County. These findings are supported by previous studies which highlight geospatial patterns of CAP admissions,^{6,8,13,24,25,28,34,42–45} environmental exposure, specifically PM_2.5, as a risk factor for CAP,^{2–5,7,13,43,46–49} and racial and socioeconomic differences in CAP distribution^44,50–52 and in exposure to air pollution.^53,54

Most pneumonia studies performed in high-income countries have not focused on socioeconomic disparities in the distribution of CAP.^35,55–58 In the United States, there have been improvements in wide health disparities in mortality from pneumonia between Black and White children.⁵⁹ In this study, we do not directly measure socioeconomic status as the small sample size increases the risk of de-identification. Instead, we use insurance and race/ethnicity as proxies. In MMA, the race/ethnicity demographic shows non-Hispanic Blacks are the majority⁶⁰ and also represent the largest racial/ethnic group living in poverty.⁶⁰ Our findings show race/ethnicity, and public insurance are associated with residence in high-risk CAP areas. This finding may imply a persisting socioeconomic disparity for CAP and is supported by previous studies.^{11,44,50,51,61–65} Potential explanations for this socioeconomic disparity include low maternal education,⁵¹ and the disparity may be potentiated by parent/patient factors such as poor access to primary care, and medication adherence issues.^62–64 This disparity is likely associated with suboptimal living conditions such as having poor indoor/outdoor air quality, as shown by high PM_2.5 exposure in high-risk CAP areas. The use of GIS to identify CAP risk areas helps demonstrate geographic high-risk CAP areas which need further study and may benefit from targeted holistic interventions including cleaner air.^44,62

The MMA has higher than average concentrations of chronic childhood respiratory conditions such as asthma which have been noted in the same geographical areas found in this study.⁶¹ This is thought to be related to the presence of industrial facilities and PM_2.5 mobile emissions sources in these areas.⁶¹ Previously published studies assessing environmental exposures and respiratory-related admissions in children have also noted CAP hotspot areas to have more sources of environmental pollution such as PM_2.5 stationary emission sites and Superfund sites.^{8,44,61,66,67} Such studies highlight the need to include geospatial risk factors of CAP in children. To the best of our knowledge, this is the first study to show the association between residence in high-risk CAP areas and exposure to high PM_2.5 using satellite-derived PM_2.5 estimates in a prospectively collected cohort of children hospitalized with radiographic and clinical CAP. This finding highlights the areas in the MMA, and other similar communities, as locales where interventions for cleaner air and other strategies to decrease air pollution can be focused and may help reduce CAP rates. Our findings suggest that residential location is associated with CAP incidence among children in the MMA.

PM_2.5 originates from the combustion process of diesel and gasoline-powered vehicles, and burning of biomass and coal to generate power among other sources.⁴⁶ Satellite-derived PM_2.5 has emerged as a more sensitive measure for assessing PM_2.5 effects on health outcomes compared with ground monitoring networks, proximity measures (see Supplementary), and use of air quality models.^14–18,68 For this study, PM_2.5 levels were obtained via high spatiotemporal resolution 1 km by 1 km grid of satellite-derived PM_2.5 concentrations. The population-derived mean for high PM_2.5 in the MMA (10.75 μg/m³) was less than the annual EPA allowable threshold (<12μg/m³). Despite this, we detected a significant and clinically relevant association of geographic clustering of CAP cases to PM_2.5. We also assessed the association using measures of PM obtained by other measures (Supplementary Table 1(b)). While a few studies have assessed respiratory outcomes in children using high resolution PM estimates,^9,10,17 this study is unique because, using satellite-derived PM_2.5, we are able to show association of possible PM_2.5 effects at exposure levels lower than the current EPA PM_2.5 limit. This contributes to the body of evidence supporting a need for continuous review and adjustment to health standards given harmful effects from PM_2.5 exposure.⁴¹

Viral infection is more likely to spread through close proximity with an infected person such as seen in daycares, homes with high number of occupants, and in children <2 years of age. The etiology of infection seen in most CAP cases residing in high-risk areas, specifically viral etiology of infection, may speak to the pattern of exposure in high-risk areas compared to low-risk areas where there may be less daycare attendance, and lower number of occupants in the home. While we did not assess daycare attendance and crowding as risk factors in this study, these are known risk factors ^20,52,69,70 and are common in areas of low socioeconomic status (SES) such as the high-risk areas in this study.

There are several limitations to our study. First, conducting a retrospective analysis on a sample that did not include prospective measurement of environmental exposures lends itself to uncertainty in the quality of data for PM. And while 1 km by 1 km PM_2.5 is thought to be highly sensitive, it is still an estimate of individual level exposure. Second, it is clear to us residents in high-risk CAP areas may be affected by many things other than pollution. There are other variables that cannot be directly measured and instead we have used proxies for such variables; for instance, we use race/ethnicity and insurance as proxies for SES. There may be other ecologic factors that may also play a role given the concentration of high-risk CAP areas in documented pollution sites (Figure 2). We recognize, given the existing literature, that hotspots tend to be present in inner city locations versus suburban areas and this is what our data and results reflect. Third, despite accounting for market share in our incidence calculations, it is unclear if market share is uniform across the entire MMA. Fourth, although local cluster detection using Turnbull’s method allows for variations in population density when adjusting for spatial clusters of CAP incidence, it requires a user-defined population size for the area. Despite this constraint, this method is computer-intensive and fully accounts for spatial autocorrelation during cluster evaluation to identify the most significant cluster locations of CAP incidence. Fifth, our sample size was small and thus may have limited power to detect differences; however, this study is useful to inform a larger prospective study of similar populations. And lastly, because this is a retrospective study using home addresses to estimate PM_2.5 exposure, we could not account for other areas (i.e. school, other homes, parks, etc.) in which the children in our population could have been exposed to indoor and outdoor pollution, which could also have been risks for the development of CAP.

In conclusion, we demonstrated that in the MMA, race/ethnicity, and insurance, both proxies of SES are significantly associated with residence in a high-risk area for CAP. In addition, spatial detection of high-risk areas for CAP in children showed these children reside in areas with higher PM_2.5 concentrations in the MMA. Interestingly, mean PM_2.5 levels of 10.75 μg/m³, but less than the annual EPA allowable threshold of <12μg/m3 were associated with high-risk areas for CAP in children suggesting that children may be particularly vulnerable to lower PM2.5 levels. The observations from this paper inform future work where the study of disease patterns with environmental influences may incorporate spatial analysis to provide a more comprehensive evaluation. Future studies could consider using personal PM monitors for children in high-risk areas to evaluate their individual exposure levels to help quantify their risk as it relates to CAP and other respiratory diseases. Lastly, the results of this study may be used to inform further research aimed at at-risk communities and vulnerable populations in an effort to reduce CAP burden.

Supplemental Material

sj-pdf-1-ebm-10.1177_15353702211014456 - Supplemental material for Particulate matter exposure predicts residence in high-risk areas for community acquired pneumonia among hospitalized children

Supplemental material, sj-pdf-1-ebm-10.1177_15353702211014456 for Particulate matter exposure predicts residence in high-risk areas for community acquired pneumonia among hospitalized children by Tonny J Oyana, Jagila Minso, Tamekia L Jones, Jonathan A McCullers, Sandra R Arnold and Stephania A Cormier in Experimental Biology and Medicine

Footnotes

AUTHORS’ CONTRIBUTIONS

JM, TJO, and SAC conceptualized the study, designed the experiments and wrote the manuscript. JM and TJO performed data collection and executed data analyses. AB, SJ, JAM, and SRA enrolled the study subjects for the EPIC study and provided the clinical and geographical data for the study. TLJ performed statistical analysis of the data and drafted statistical methods. All authors reviewed and approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Declaration OF CONFLICTTING INTERESTS

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

FUNDING

The Etiology of Pneumonia in the Community (EPIC) study was supported by the Influenza Division of the National Center for Immunization and Respiratory Diseases at the Centers for Disease Control through cooperative agreements with each study site. This publication received support from the Le Bonheur Children’s Foundation Grant to JMW. It was also supported by the National Institutes of Health (NIEHS P42ES013648 and R01ES015050 and NIAID R01AI090059) to SAC. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

DISCLAIMER

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.

ORCID iD

Stephania A Cormier

SupplementAL MATERIAL

Supplemental material for this article is available online.

References

Bryce

Boschi-Pinto

Shibuya

Black

Group

WHO estimates of the causes of death in children. Lancet 2005; 365:1147–52

Darrow

Klein

Flanders

Mulholland

Tolbert

Strickland

MJ.

Air pollution and acute respiratory infections among children 0-4 years of age: an 18-year time-series study. Am J Epidemiol 2014; 180:968–77

Mehta

Shin

Burnett

North

Cohen

AJ.

Ambient particulate air pollution and acute lower respiratory infections: a systematic review and implications for estimating the global burden of disease. Air Qual Atmos Health 2013; 6:69–83

Ostro

Roth

Malig

Marty

The effects of fine particle components on respiratory hospital admissions in children. Environ Health Perspect 2009; 117:475–80

Hertz-Picciotto

Baker

Yaw

Dostal

Joad

Lipsett

Greenfield

Herr

CEW

Benes

Shumway

Pinkerton

Sram

Early childhood lower respiratory illness and air pollution. Environ Health Perspect 2007; 115:1510–8

MacIntyre

Gehring

Molter

Fuertes

Klumper

Kramer

Quass

Hoffmann

Gascon

Brunekreef

Koppelman

Beelen

Hoek

Birk

de Jongste

Smit

Cyrys

Gruzieva

Korek

Bergstrom

Agius

de Vocht

Simpson

Porta

Forastiere

Badaloni

Cesaroni

Esplugues

Fernandez-Somoano

Lerxundi

Sunyer

Cirach

Nieuwenhuijsen

Pershagen

Heinrich

Air pollution and respiratory infections during early childhood: an analysis of 10 European birth cohorts within the ESCAPE project. Environ Health Perspect 2014; 122:107–13

Anderson

Thundiyil

Stolbach

Clearing the air: a review of the effects of particulate matter air pollution on human health. J Med Toxicol 2012; 8:166–75

Hruba

Fabianova

Koppova

Vandenberg

JJ.

Childhood respiratory symptoms, hospital admissions, and long-term exposure to airborne particulate matter. J Expo Sci Environ Epidemiol 2001; 11:33–40

Liu

Mong

Chou

CCK.

Spatial correlation of satellite-derived PM2.5 with hospital admissions for respiratory diseases. Remote Sens 2016; 8:914

10.

Strickland

Hao

Chang

Darrow

Liu

Pediatric emergency visits and short-term changes in PM2.5 concentrations in the U.S. State of Georgia. Environ Health Perspect 2016; 124:690–6

11.

Yap

Gilbreath

Garcia

Jareen

Goodrich

The influence of socioeconomic markers on the association between fine particulate matter and hospital admissions for respiratory conditions among children. Am J Public Health 2013; 103:695–702

12.

Nascimento

LFC

Rizol

PMSR

Peneluppi

AP.

Estimating the average length of hospitalization due to pneumonia: a fuzzy approach. Braz J Med Biol Res 2014; 47:977–81

13.

Cesar

Nascimento

Mantovani

Pompeo

VL.

Fine particulate matter estimated by mathematical model and hospitalizations for pneumonia and asthma in children. Rev Paul Pediatr 2016; 34:18–23

14.

van Donkelaar

Martin

Brauer

Boys

BL.

Use of satellite observations for long-term exposure assessment of global concentrations of fine particulate matter. Environ Health Perspect 2015; 123:135–43

15.

van Donkelaar

Martin

Spurr

Burnett

RT.

High-resolution satellite-derived P. M25 from optimal estimation and geographically weighted regression over North America. Environ Sci Technol 2015; 49:10482–91

16.

Lee

Kloog

Chudnovsky

Lyapustin

Wang

Melly

Coull

Koutrakis

Schwartz

Spatiotemporal prediction of fine particulate matter using high-resolution satellite images in the southeastern US 2003-2011. J Expo Sci Environ Epidemiol 2016; 26:377–84

17.

Lin

Yao

Fung

Lau

AK.

Assessment of health burden caused by particulate matter in Southern China using high-resolution satellite observation. Environ Int 2017; 98:160–70

18.

Waller

Lyapustin

Wang

Al-Hamdan

Crosson

Estes

Quattrochi

Puttaswamy

Liu

Estimating ground-level PM2.5 concentrations in the southeastern United States using MAIAC AOD retrievals and a two-stage model. Remote Sens Environ 2014; 140:220–32

19.

Heiskanen-Kosma

Korppi

Jokinen

Heinonen

Risk factors for community-acquired pneumonia in children: a population-based case-control study. Scand J Infect Dis 1997; 29:281–5

20.

Fonseca Lima

Mello

Albuquerque

Lopes

Serra

Lima

Correia

JB.

Risk factors for community-acquired pneumonia in children under five years of age in the post-pneumococcal conjugate vaccine era in Brazil: a case control study. BMC Pediatr 2016; 16:157

21.

Ding

Zhang

Zhong

Xiao

Zhou

Duan

Xie

Gao

HC.

Viral etiology and risk factors for severe community-acquired pneumonia in children. Zhongguo Dang Dai Er Ke Za Zhi 2012; 14:449–53

22.

Grant

Emery

Milne

Coster

Forrest

Wall

Scragg

Aickin

Crengle

Leversha

Tukuitonga

Robinson

EM.

Risk factors for community-acquired pneumonia in pre-school-aged children. J Paediatr Child Health 2012; 48:402–12

23.

Blain

Thomas

Shirley

Simmister

Elemraid

Gorton

Pearce

Clark

Rushton

Spencer

, North East of England Paediatric Respiratory Infection Study Group. Spatial variation in the risk of hospitalization with childhood pneumonia and empyema in the North of England. Epidemiol Infect 2014; 142:388–98

24.

Andrade

Silva

Martelli

Oliveira

Morais Neto

Siqueira

Melo

FJ.

Population-based surveillance of pediatric pneumonia: use of spatial analysis in an urban area of Central Brazil. Cad Saúde Pública 2004; 20:411–21

25.

Mukai

Alves

KDC

Nascimento

LFC.

Spatial analysis of hospitalizations for pneumonia in the Vale do Paraiba region of Brazil. J Bras Pneumol 2009; 35:753–8

26.

Ranalli

Rocco

Jona Lasinio

Moroni

Castellini

Crocchianti

Cappelletti

Functional exploratory data analysis for high-resolution measurements of urban particulate matter. Biom J 2016; 58:1229–47

27.

Gorton

Jones

JL.

Wide geographic variation between Pennsylvania counties in the population rates of hospital admissions for pneumonia among children with and without comorbid chronic conditions. Pediatrics 2006; 117:E176–E80

28.

Crighton

Elliott

Moineddin

Kanaroglou

Upshur

RE.

An exploratory spatial analysis of pneumonia and influenza hospitalizations in Ontario by age and gender. Epidemiol Infect 2007; 135:253–61

29.

Clarke

McLafferty

Tempalski

BJ.

On epidemiology and geographic information systems: a review and discussion of future directions. Emerg Infect Dis 1996; 2:85–92

30.

Musa

Chiang

Sylk

Bavley

Keating

Lakew

Tsou

Hoven

CW.

Use of GIS mapping as a public health tool-from cholera to cancer. Health Serv Insights 2013; 6:111–6

31.

Osei

Duker

AA.

Spatial dependency of V. cholera prevalence on open space refuse dumps in Kumasi, Ghana: a spatial statistical modelling. Int J Health Geogr 2008; 7:62

32.

Liu

Wang

Liu

Sun

Ding

Zhang

Xue

Detecting spatial-temporal clusters of HFMD from 2007 to 2011 in Shandong province, China. PLoS One 2013; 8:e63447

33.

Samphutthanon

Tripathi

Ninsawat

Duboz

Spatio-temporal distribution and hotspots of hand, foot and mouth disease (HFMD) in Northern Thailand. Int J Environ Res Public Health 2013; 11:312–36

34.

Nascimento

LFC

de Medeiros

APP.

Admissions due to pneumonia and biomass burning: a spatial approach. J Pediat-Brazil 2012; 88:177–83

35.

Jain

Williams

Arnold

Ampofo

Bramley

Reed

Stockmann

Anderson

Grijalva

Self

Zhu

Patel

Hymas

Chappell

Kaufman

Kan

Dansie

Lenny

Hillyard

Haynes

Levine

Lindstrom

Winchell

Katz

Erdman

Schneider

Hicks

Wunderink

Edwards

Pavia

McCullers

Finelli

Team

CES.

Community-acquired pneumonia requiring hospitalization among US children. N Engl J Med 2015; 372:835–45

36.

Yasaitis

Arcaya

Subramanian

SV.

Comparison of estimation methods for creating small area rates of acute myocardial infarction among medicare beneficiaries in California. Health Place 2015; 35:95–104

37.

Cressie

Smoothing regional maps using empirical Bayes predictors. Geograph Anal 1992; 24:75–95

38.

Turnbull

Iwano

Burnett

Howe

Clark

LC.

Monitoring for clusters of disease – application to leukemia incidence in upstate New-York. Am J Epidemiol 1990; 132:S136–43

39.

Newcombe

RG.

Two-sided confidence intervals for the single proportion: comparison of seven methods. Statist Med 1998; 17:857–72

40.

Wilson

EB.

Probable inference, the law of succession, and statistical inference. J Am Stat Assoc 1927; 22:209–12

41.

Guidotti

TL.

Decision time on standards for particulate matter in the United States. Arch Environ Occup Health 2013;63–5

42.

Crighton

Elliott

Kanaroglou

Moineddin

Upshur

RE.

Spatio-temporal analysis of pneumonia and influenza hospitalizations in Ontario. Geospat Health 2008; 2:191–202

43.

Cesar

Nascimento

Carvalho

Jr.

Association between exposure to particulate matter and hospital admissions for respiratory disease in children. Rev Saude Publica 2013; 47:1209–12

44.

Thorn

Minamisava

Nouer

Ribeiro

Andrade

AL.

Pneumonia and poverty: a prospective population-based study among children in Brazil. BMC Infect Dis 2011; 11:180

45.

Morris

Munasinghe

RL.

Geographic variability in hospital admission rates for respiratory disease among the elderly in the United States. Chest 1994; 106:1172–81

46.

Nascimento

LFC

Vieira

LCPF

Mantovani

KCC

Moreira

DS.

Air pollution and respiratory diseases: ecological time series. Sao Paulo Med J 2016; 134:315–21

47.

Barnett

Williams

Schwartz

Neller

Best

Petroeschevsky

Simpson

RW.

Air pollution and child respiratory health: a case-crossover study in Australia and New Zealand. Am J Respir Crit Care Med 2005; 171:1272–8

48.

DeMeo

Zanobetti

Litonjua

Coull

Schwartz

Gold

DR.

Ambient air pollution and oxygen saturation. Am J Respir Crit Care Med 2004; 170:383–7

49.

Shah

Ramankutty

Premila

Sathy

Risk factors for severe pneumonia in children in South Kerala: a hospital-based case-control study. J Trop Pediatr 1994; 40:201–6

50.

Henrickson

Kuhn

Savatski

LL.

Epidemiology and cost of infection with human parainfluenza virus type-1 and type-2 in young-children. Clin Infect Dis 1994; 18:770–9

51.

Azab

Sherief

Saleh

Elsaeed

Elshafie

Abdelsalam

SM.

Impact of the socioeconomic status on the severity and outcome of community-acquired pneumonia among Egyptian children: a cohort study. Infect Dis Poverty 2014; 3:14

52.

Victora

Fuchs

Flores

Fonseca

Kirkwood

Risk factors for pneumonia among children in a Brazilian metropolitan area. Pediatrics 1994; 93:977–85

53.

Stewart

Mitchell

Edgerton

VanCott

Environmental justice and health effects of urban air pollution. J Natl Med Assoc 2015; 107:50–8

54.

Hajat

Hsia

O'Neill

MS.

Socioeconomic disparities and air pollution exposure: a global review. Curr Environ Health Rep 2015; 2:440–50

55.

Michelow

Olsen

Lozano

Rollins

Duffy

Ziegler

Kauppila

Leinonen

McCracken

Jr.

Epidemiology and clinical characteristics of community-acquired pneumonia in hospitalized children. Pediatrics 2004; 113:701–7

56.

Madhi

De Wals

Grijalva

Grimwood

Grossman

Ishiwada

Lee

Nascimento-Carvalho

Nohynek

O'Brien

Vergison

Wolter

The burden of childhood pneumonia in the developed world: a review of the literature. Pediatr Infect Dis J 2013; 32:e119–27

57.

Chang

Bednarczyk

Becker

Hockenberry

Weiss

Orenstein

Omer

SB.

Trends in U.S. hospitalizations and inpatient deaths from pneumonia and influenza, 1996-2011. Vaccine 2016; 34:486–94

58.

Griffin

Zhu

Moore

Whitney

Grijalva

CG.

U.S. hospitalizations for pneumonia after a decade of pneumococcal vaccination. N Engl J Med 2013; 369:155–63

59.

Dowell

Kupronis

Zell

Shay

DK.

Mortality from pneumonia in children in the United States, 1939 through 1996. N Engl J Med 2000; 342:1399–407

60.

USA D. p. Data USA: Memphis, TN-MS-AR Metro Area, 2017. Available from: https://datausa.io/profile/geo/memphis-tn-ms-ar-metro-area/#economy

61.

Oyana

Podila

Wesley

Lomnicki

Cormier

Spatiotemporal patterns of childhood asthma hospitalization and utilization in Memphis metropolitan area from 2005 to 2015. J Asthma 2017; 54:842–55

62.

Bradley

Byington

Shah

Alverson

Carter

Harrison

Kaplan

Mace

McCracken

Moore

St Peter

Stockwell

Swanson

JT.

The management of community-acquired pneumonia in infants and children older than 3 months of age: clinical practice guidelines by the pediatric infectious diseases society and the Infectious Diseases Society of America. Clin Infect Dis 2011; 53:E25–E76

63.

Agha

Glazier

Guttmann

Relationship between social inequalities and ambulatory care-sensitive hospitalizations persists for up to 9 years among children born in a major Canadian urban center. Ambul Pediatr 2007; 7:258–62

64.

Flores

Abreu

Chaisson

Sun

DL.

Keeping children out of hospitals: parents' and physicians' perspectives on how pediatric hospitalizations for ambulatory care-sensitive conditions can be avoided. Pediatrics 2003; 112:1021–30

65.

Cohen

Social status and susceptibility to respiratory infections. Ann N Y Acad Sci 1999; 896:246–53

66.

Brauer

Hoek

Smit

de Jongste

Gerritsen

Postma

Kerkhof

Brunekreef

Air pollution and development of asthma, allergy and infections in a birth cohort. Eur Respir J 2007; 29:879–88

67.

Brauer

Hoek

Van Vliet

Meliefste

Fischer

Wijga

Koopman

Neijens

Gerritsen

Kerkhof

Heinrich

Bellander

Brunekreef

Air pollution from traffic and the development of respiratory infections and asthmatic and allergic symptoms in children. Am J Respir Crit Care Med 2002; 166:1092–8

68.

Sayer

Levy

Zhang

Xue

Tong

Huang

Liu

Satellite-based spatiotemporal trends in PM2.5 concentrations: China, 2004-2013. Environ Health Persp 2016; 124:184–92

69.

Jackson

Mathews

Pulanic

Falconer

Rudan

Campbell

Nair

Risk factors for severe acute lower respiratory infections in children: a systematic review and meta-analysis. Croat Med J 2013; 54:110–21

70.

Simoes

EA.

Environmental and demographic risk factors for respiratory syncytial virus lower respiratory tract disease. J Pediatr 2003; 143:S118–26

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.28 MB