Associations among neighborhood greenspace, neighborhood violence, and children’s asthma control in an urban city

Among 6.1 million children with asthma,¹ 38% have uncontrolled asthma leading to increased pediatric emergency department (ED) visits, hospitalizations, and healthcare costs.² Uncontrolled asthma unequally harms children from lower-income, racial and ethnic minority families.¹ Disparities in asthma control have been attributed to differential exposures to poverty, living in unsafe, stressful neighborhoods, and unhealthy physical environments.³ One modifiable factor that may improve asthma control is the availability of neighborhood greenspace, i.e., land with grass, trees, or other vegetation. Greenspace is associated with increased physical activity and decreased stress, environmental heat, air pollution and disproportionately benefits white, affluent neighborhoods.⁴ Although these variables are associated with improved pediatric asthma control, testing relationships between greenspace and asthma control has shown mixed results,⁵ potentially due to failure to account for neighborhood level factors that influence children’s exposure to greenspace, such as neighborhood violence. Living in an unsafe neighborhood is associated with increased stress and spending less time outside, resulting in increased exposure to indoor allergens and secondhand smoke (SHS). Children living in areas with higher violent crime have poorer asthma control than those in areas with lower crime rates.⁶ Our study examined whether the availability of neighborhood greenspace was associated with improved asthma control and if neighborhood violence might modify that relationship by limiting children’s exposure to greenspace.

This cross-sectional study used data from a larger study of predominantly African American, Medicaid insured children with persistent asthma recruited from an urban pediatric ED following an asthma exacerbation.⁷ It was approved by the Johns Hopkins Hospital IRB. At baseline we collected demographic and social risk data, SHS exposure, asthma medication use, and child blood and saliva samples to determine allergen sensitization to indoor and outdoor allergens and SHS exposure. Medication use was based on the asthma medication ratio (AMR). The child’s home address was geocoded to neighborhood-level variables of greenspace and violence.⁸

Greenspace was measured using the normalized difference vegetation index (NDVI), an index from −1 to 1 where higher numbers indicate greener vegetation areas, calculated using satellite data. Violence was measured using the violent-crime-victimization rate calculated from police reported crime data. Level of asthma control, categorized as not well controlled (NWC) or very poorly controlled (VPC; the worse outcome), was based on parents’ report of their child’s asthma symptom days, symptom nights, activity limitation, and rescue inhaler use. For missing data on AMR and social risk (11% and 8%, respectively), we used the Markov Chain Monte Carlo method of multiple imputation with 10 data sets. Hypotheses were tested using logistic regression and STATA version 15.1 software.

The sample (n=196 children) was predominantly male (65%), African American (95%), and Medicaid-insured (95%). Most children had VPC asthma (55%), SHS exposure (59%), and sensitization to indoor and outdoor allergens (83% and 63%, respectively). Most caregivers were single (78%) and the child’s biological mother (93%). NDVI values ranged from 0.00 to 0.17 (M=0.0, SD=0.04). Violent-crime-victimization rate ranged from 4.94 to 52.23 violent-crime-victimizations per 1,000 census tract residents (M=20.75,SD=9.74).

Multivariable logistic regression results indicate that greenspace was not associated with asthma control (Table 1-Model 1). However, AMR was associated with asthma control (aOR=2.41, 95%CI=1.18,4.94). Furthermore, violence was not associated with the level of asthma control but AMR remained associated with asthma control (aOR=2.41, 95%CI=1.18,4.96) (Model 2). An interaction term between violence and greenspace was added into a logistic regression model to test moderation. As shown (Model 3), violence did not moderate the relationship between greenspace and the level of asthma control. In sensitivity analyses controlling for reported length of time in the neighborhood, caregiver age, and number of allergen sensitizations results were unchanged.

Table 1.

Logistic Regression examining the odds of Very Poorly Controlled (VPC) Asthma (n = 167)

	Odds ratio (95% CI)
Variable	Unadjusted relationships OR (95% CI)	Model 1 aOR (95% CI)	Model 2 aOR (95% CI)	Model 3 aOR (95% CI)
Child age	1.06 (0.95, 1.18)
Child sex (male)	1.25 (0.21, 7.68)
Social Risk Indexa	1.13 (0.93, 1.37)	1.18 (0.93, 1.50)	1.18 (0.92, 1.51)	1.84 (0.93, 1.51)
Season @ Baseline: Fall	Ref	Ref	Ref	Ref
Winter	1.69 (0.81, 3.51)	1.25 (0.54, 2.92)	1.26 (0.54, 2.93)	1.26 (0.54, 2.94)
Spring	0.92 (0.42, 2.02)	0.96 (0.40, 2.27)	0.95 (0.40, 2.25)	0.97 (0.41, 2.29)
Summer	1.23 (0.54, 2.80)	0.81 (0.31, 2.12)	0.81 (0.31, 2.13)	0.81 (0.31, 2.14)
Asthma Medication Ratio ≥0.50	2.20 (1.20, 4.05)*	2.41 (1.18, 4.94)*	2.41 (1.18, 4.96)*	2.45 (1.18, 5.10)*
Indoor Allergen Sensitizationb (+)	0.60 (0.27, 1.36)
Outdoor Allergen Sensitizationc (+)	0.64 (0.35, 1.18)	0.56 (0.29, 1.11)	0.56 (0.28, 1.11)	0.56 (0.28, 1.10)
Secondhand Smoke Exposured (+)	0.72 (0.39, 1.31)	0.55 (0.27, 1.10)	0.55 (0.27, 1.11)	0.54 (0.27, 1.09)
Greenspace	1.04 (0.59, 1.83)	1.01 (0.93, 1.10)		0.98 (0.79, 1.21)
Violencee	1.01 (0.98, 1.04)		1.00 (0.97, 1.03)	0.99 (0.91, 1.07)
Interactions
Violence X Greenspace				1.00 (0.99, 1.01)

^aSocial Risk Index includes: high school graduate/GED or less, single-parent household, Medicaid insured, unemployed, ≥4 children living in the home, parent Centers for Epidemiological Studies Depression score ≥16, if they have lived ≤ 1 year in their current home, and if they perceived their neighborhood as having a weak sense of community, scores range from 0–8 with higher scores indicating greater risk.

^bIndoor allergen sensitization + if IgE antibody ≥0.35 kU/L to ≥1 indoor allergen (mouse, cockroach, cat, dog, Aspergillus mold, or house dust mite).

^cOutdoor allergen sensitization + if IgE antibody ≥0.35 kU/L to ≥1 outdoor allergen (timothy grass, Alternaria mold, oak tree, or common ragweed).

^dSecondhand smoke exposure + (salivary cotinine ≥1.00 ng/ml).

^eViolence measured as violent-crime-victimization rate per 1,000 census tract residents.

^*p<0.05

Our data indicate that the availability of greenspace was not related to children’s asthma control, and neighborhood violence did not moderate this relationship. These null results align with a prior Australian study failing to find associations between the interaction of greenspace and area safety on children’s asthma diagnosis, despite using different populations and measures for greenspace and violence.⁹

There are several study limitations. While NDVI is an objective and validated measure of neighborhood greenspace and measures quantity of green vegetation, NDVI does not specifically measure characteristics of greenspace associated with improved child health. For example, NDVI does not distinguish between a vacant lot of weeds and a neighborhood park that may encourage physical activity and psychological restoration, pathways through which greenspace improves health. We lacked data on children’s actual exposure to greenspace, which might have occurred in a different geographic area such as a friend’s home or at school. Asking the location and duration of outdoor time may provide a more comprehensive estimate of greenspace exposure.

Using existing data constrained the sample size, sample variability, and number of variables for examination. Although a power analysis demonstrated the sample size was sufficient to test our primary hypothesis, it was insufficient to test moderation. Additionally, all participants had uncontrolled asthma, further limiting our power to detect differences between children with NWC and VPC asthma. We purposefully recruited children with uncontrolled asthma since they experience the highest asthma burden. However, this diminished variability may have biased results toward the null.

Despite these study limitations, multiple methods and informants (geocoding, crime statistics, and physiologic measures) were used to understand how neighborhood factors are linked to asthma control in high-risk children. The study harnessed satellite data to calculate NDVI, and included neighborhood violence as a contextual factor to determine potential benefits of neighborhood greenspace for children with asthma.

Results highlight the complexities of studying the relationships between neighborhood factors and asthma control in high-risk children. Researchers previously demonstrated that elements of neighborhood context, e.g., housing-code violations, were associated with increased allergen exposures potentially increasing asthma morbidity.¹⁰ Similar to housing-code violations, greenspaces are not randomly located but rather co-located with other environmental and social factors. Further research in this area should collect primary, longitudinal data using a larger sample of children with more variability in asthma control levels. Effective ways to modify neighborhood environments that are toxic to children’s health, particularly for children with asthma, need to be explored.

Abbreviations

AMR

Asthma Medication Ratio

NDVI

Normalized difference vegetation index

NWC

Not well controlled

OR

odds ratio

PED

Pediatric Emergency Department

SHS

Secondhand smoke

VPC

Very poorly controlled