The novel objective of this study was to use repeated measures of GIS indicators of residential proximity and density of traffic and stationary sources of air pollution to represent long-term air pollution exposures of young children living in high-density New York City neighborhoods in Northern Manhattan and South Bronx. In analyses of repeated measures between birth and age 5 years, density of four-way intersections was associated significantly with wheeze, and proximity to highways was associated significantly with total IgE levels. Additionally, the percentage of commercial building area was associated positively with asthma, wheeze, and total IgE. The positive association between proximity to stationary sources of air pollution and reported asthma approached statistical significance. Moreover, in longitudinal analyses, several GIS variables assessed at the prenatal or earlier childhood residence were associated with wheeze and asthma later in childhood, particularly among subjects who never moved. Given that GIS traffic indicators have been strong predictors of ambient air pollutant concentrations across diverse urban locations, these findings indicate that concurrent, prenatal, early childhood, and cumulative exposure to air pollution each may contribute to respiratory morbidity among children during the first five years of life.
Approximately 50% of our cohort resided within 400-480 m of a state or county highway, a zone of elevated concentrations of traffic-related pollutants (Zhou and Levy 2007
), as do 75% and 15% of children residing in Manhattan and the Bronx, respectively (Environmental Defense Fund, 2007
). While the present findings may be limited to this cohort of Dominican and African American children, given the high percentage of urban populations residing near major roadways and the high prevalence of urban asthma (Akinbami et al., 2011
), the implications of these findings for public health may be quite broad.
The major strength of this study was its longitudinal design. To date, most GIS-based studies have used GIS variables ascertained at a single point in time to represent long-term average exposures (Chang et al., 2009
; Clark et al., 2010
; Gordian et al., 2006
; McConnell et al., 2006
; Ryan et al., 2007
). An additional strength was the availability of data from most subjects for 3-4 time points. Combined, these approaches allowed us to discern more accurately the temporality of associations between exposure and health outcomes and account for changing exposures in a population that moves frequently. Further, with these highly time-resolved GIS data, we were able to compare associations among different time windows of exposure. Such comparisons are important given prior evidence in this cohort of New York City children that both concurrent and prenatal air pollution exposures are associated with recurrent respiratory symptoms and asthma (Miller et al., 2004
; Patel et al., 2009a
; Rosa et al., 2010
Because we did not examine associations between the various GIS variables and concentrations of ambient air pollutants, the findings may not be attributable entirely to the effects of exposure to particular traffic-related air pollutants but also may reflect effects related to socioeconomic status, social stressors such as demoralization or violence, or indoor sources of toxicants. However, in studies across diverse locations, different indicators of traffic and built environment have been associated with pollutants such as NO2
, and black carbon (Brauer et al., 2008
; Morgenstern et al., 2008
; Patel et al., 2009b
). By examining a diverse set of GIS variables representing traffic proximity, density, vehicle mix, industrial emission sources, and land use characteristics, we aimed to characterize whether particular surrogates of local air pollution mix and perhaps exposures to a mixture of air pollutants were associated with different measures of respiratory health. Further, the yearly individual-level GIS data provided information on air pollution exposures with improved spatial and temporal resolution compared with the air pollutant measurements available from the one to two central monitoring sites.
Associations of GIS indicators varied across the outcomes of wheeze, asthma, and IgE. This result was expected because the relationships among these measures are not fixed early in childhood, especially before age 5 or 6 years (Martinez 2002
). Transient wheeze often due to viral infection is common in infancy (Brooks and Lemanske 2002
; Lin et al., 2007
) and may not predict persistent asthma later in childhood. Atopy also becomes more strongly associated with persistent asthma at older ages (Martinez 2002
). Consistent with these published reports, the proportion of children in this cohort with wheeze was highest at age 1 year, whereas the proportion of children reported to be diagnosed with asthma was highest at age 5 years. At each age, there were children with reported asthma who did not wheeze in the previous 12 months (32%, 44%, 47%, and 18% at ages 1, 2, 3, and 5 years, respectively) and children without asthma who did wheeze (31%, 13%, 10%, and 12% at ages 1, 2, 3, and 5 years, respectively).
In this study, asthma was ascertained by parental report of a physician diagnosis. Diagnostic criteria were likely to vary among subjects’ physicians. Additionally, there are limitations to asthma diagnoses in children before age 5 years. Other conditions may have similar symptoms (Pedersen et al., 2011
), and asthma before age 5 years may be transient (Martinez 2002
). Despite these issues, we expect any error in the ascertainment of asthma to be independent of exposure to air pollution sources and more likely would have resulted in the underestimation of the effects of exposure to traffic and stationary sources of air pollution.
Notably, proximity to highway was associated not only with wheeze and total IgE but also with reported asthma among subjects who never moved, indicating that higher cumulative exposures to traffic-related pollutants between birth and age 5 years may be associated with all three outcomes. In support of our findings, other studies of children in the same age range also have found that greater proximity to major roadway (McConnell et al., 2006
) or higher ambient air concentrations of traffic-related pollutants are associated with asthma among children who never moved (Gehring et al., 2010
). Also in support of a cumulative exposure effect, we found that multiple measures of traffic volume at the age 1 residence, including proximity to highway, four-way intersection density, and roadway density, were associated significantly with wheeze at age 5 years in subjects who never moved.
The positive association between proximity to highway and total IgE is supported by previous findings that ambient concentrations of traffic-related pollutants such as nitrogen oxides and polycyclic aromatic hydrocarbons are associated with total IgE in children (Herr et al., 2011
; Janssen et al. 2003
). It has been postulated that ambient traffic-related particles can carry adsorbed allergens into airways and serve as adjuvants during the development of allergy (Diaz-Sanchez et al., 1994
; Knox et al., 1997
; Liu et al., 2008
; Takenaka et al., 1995
). We did not find GIS variables to be associated with sensitization to indoor allergens. However, evidence suggests that traffic density measures and air pollution exposures are associated with sensitization to outdoor allergens (Janssen et al., 2003
; Kramer et al., 2000
; Mortimer et al., 2008
) and even food allergens (Brauer et al., 2007
) rather than indoor allergens. Thus, the effect of traffic-related air pollution exposure on IgE may be dependent on specific allergens, and in the present study, we only had data on sensitization to indoor allergens.
Among the GIS indicators related specifically to diesel vehicle emissions (e.g., the number of bus stops, truck route density), we found the number of bus stops to be associated significantly with asthma but only when examining prenatal exposure or concurrent exposure among subjects who never moved. Characterizing associations with diesel emissions sources is difficult in the present study because of changes made to the composition of the New York City transit bus fleet (26% diesel-electric hybrid, 27% compressed natural gas, 95% of diesel buses fit with particulate filters by the end of the study period) and because of varying truck volumes among truck routes (Metropolitan Transit Authority, 2011
; New York State Department of Transportation, 2011
). The exposure measurement error associated with variable contributions of bus and truck emissions to ambient air concentrations of diesel exhaust particles across space and time may have biased associations with respiratory outcomes to the null. Additional investigation is required to characterize associations of the number of bus stops and truck route density with elemental or black carbon air concentrations in the present study area and whether these associations change over time.
We acknowledge that re-examination of these associations using smaller buffer sizes may have been informative. Meta-analyses have indicated that the concentrations of traffic-related pollutants remain elevated over a range of 100 to 500 m (Karner et al., 2011; Zhou and Levy 2007
), a range that includes our 250 m buffer size. Additionally, several studies have found asthma or other respiratory outcomes in children to be associated with GIS variables calculated with a similar buffer size or pollutant concentrations estimated with similar buffer sizes (Brauer et al., 2007
; Chang et al., 2009
; Clark et al., 2010
; Gehring et al., 2010
; Maantay 2007
; Morgenstern et al. 2007
; Ryan et al., 2007
). Associations also have been found for buffer sizes in the range of 50 to 100 m (Brauer et al., 2007
; Clougherty et al., 2008
; Gilbert et al., 2005
; Gordian et al., 2006
; McConnell et al., 2010
; Ryan et al., 2007
). Thus, in this study, we may have underestimated associations of some traffic sources with respiratory outcomes.
Associations of proximity to stationary sources of air pollution with respiratory morbidity in young children have not been examined widely. In the present study, we assessed the proportion of the 250 m radial buffer located within 0.80 km of an industrial facility included in the 2005 National Emissions Inventory or Toxic Release Inventory. No sites are located in Manhattan; however, facilities with emissions of polycyclic aromatic and other hydrocarbons and lead are located in the South Bronx (U.S. Environmental Protection Agency, 2005
). The 0.80 km buffer was designated because it represents an area heavily impacted by source emissions and is associated with asthma exacerbations in the Bronx (Maantay 2007
). The associations with asthma that we observed among subjects who moved and for age 2 exposures build on previous cross-sectional evidence (Clark et al., 2010
; Maantay 2007
) by suggesting that exposures earlier in childhood may impact the subsequent development of asthma. Commercial building area also has not been examined widely as an indicator of air pollution exposure. In the present study, commercial building area was associated positively with wheeze, asthma, and IgE. Commercial building area may serve as a good proxy for the level of traffic activity, and presumed traffic emissions, near the residence. Notably, commercial building area was correlated weakly with most other evaluated GIS variables and thus, also may serve as a surrogate for pollution mixtures or other complex characteristics of the urban environment, such as socioeconomic status and housing conditions that are associated with asthma and allergy.
Roadway density and proximity to highway at the prenatal residence were associated with reported asthma. Both prenatal and concurrent proximity to highway were associated more strongly with asthma among subjects who never moved, indicating that cumulative exposure may be more important than prenatal exposure alone. In contrast, the association of prenatal roadway density with asthma was significant only among children who moved after the prenatal period. While the critical time window of exposure may vary among different air pollution sources and the particular pollutants they may represent, it is important to acknowledge the difficulties in separating prenatal from concurrent exposures because repeated measures of GIS variables were highly correlated within subjects. In a recent study of this cohort, in main effects analyses, prenatal exposures to polycyclic aromatic hydrocarbons were not associated with asthma or IgE after controlling for postnatal exposures (Rosa et al., 2010
). Further investigation of associations between GIS variables and residential concentrations of specific air pollutants will permit more meaningful comparisons across the various studies conducted in the Columbia Center for Children’s Environmental Health cohort.