Predictors of Depressive Symptoms in Caregivers of Children With Poorly Controlled Asthma

Abstract

Children residing in low-income neighborhoods are disproportionately affected by asthma morbidity and mortality. Neighborhood violence has been explored in relationship to child morbidity and health and developmental outcomes, but less is known about the relationship of violence to caregiver mental health. The purpose of this study was to examine the relationship of neighborhood violent crime victimization (objective and subjective measures), perceptions of community well-being and support, and depressive symptoms among a sample of primarily single female caregivers of children with uncontrolled asthma. This is a secondary analysis of baseline data obtained from a randomized controlled trial of a home-based environmental control intervention for children aged 3 to 12 years, who were primarily African American, and diagnosed with persistent, uncontrolled asthma. Results showed that both objective and subjective measures of crime, particularly in those with relatively low life stress (P < .001), limited education of the caregiver (P < .001), and fewer children (P < .01) in the household had direct associations with depressive symptoms in caregivers of children with uncontrolled asthma. Neighborhood perceptions of satisfaction and a sense of community, as well as perceptions of social support, were not associated with depressive symptoms. Our findings emphasize the need to screen for depressive symptoms, life stress, as well as both objective and subjective perceptions of neighborhood violence among caregivers of children with poorly controlled asthma. Furthermore, when providing holistic care to these caregivers, stress reduction and the provision of mental health resources are paramount.

The disproportionate burden of asthma among children from low-income families and racial and ethnic minority groups is well documented.^1–3 From 2000 to 2010, these disparities increased, with Black children twice as likely to be diagnosed with asthma compared with non-Hispanic white children.⁴ In 2017, the prevalence of asthma was higher among Black children (12.6%) than among non-Hispanic white children (7.7%).⁵ These asthma disparities are not due to children being from different races or ethnic groups but from their experiences of being impacted by factors related to racial and economic oppression, including inequities in their exposure to physical and social determinants of health.⁶ Physical environmental factors, such as air pollution or secondhand smoke, and social environmental factors, such as urban versus rural settings, are increasingly recognized as important contributors to the observed increased risk for asthma morbidity and severity among children from racial and ethnic minority populations.^6–8 Those in poverty are more likely to live in neighborhoods with substandard housing and higher rates of unemployment, which are also associated with asthma morbidity.⁶ Children and families from racial and ethnic minority groups have differential exposures to neighborhood factors. These factors may be due to historical and social influences, resulting in segregated communities that are often disinvested, resulting in unique residential disadvantages with high levels of poverty and other problems.^9,10

More than a decade ago, Wright and colleagues^11,12 described the relationship of socially toxic neighborhoods to both poor health and mental health outcomes. Community violence, for example, was specifically identified as a chief social/environmental factor negatively impacting child asthma management and outcomes including increased asthma day and night symptoms.^11–17 One potential link between the negative influences of violence on child asthma outcomes is caregiver psychological distress. Caregivers impacted by chronic stress via living within violent neighborhoods often experience depressive symptoms, which, in turn, negatively impact the health of their children.^18–23 For example, maternal depressive symptoms are related to increased child asthma morbidity and health care utilization.²³ Caregivers who endorse high levels of depressive symptoms are 30% more likely to report an emergency department (ED) visit for their child than caregivers with low depressive symptomatology.²⁴ One prospective study found that baseline maternal depressive symptoms predicted childhood asthma symptoms 6 months later in low-income African American families in urban areas.¹⁹

While community violence is associated with psychological distress, there are certain gaps in our understanding of this relationship.^11,25 Although several noteworthy studies have attempted to tease apart the relationship between caregiver mental health and community violence, less is known about the relative associations between objective and subjective measures of violence and caregiver mental health.^15,26,27 Curry and colleagues²⁶ examined crime at the block level to determine whether neighborhood crime was directly associated with depressive symptoms in Baltimore, Maryland. Tonorezos and colleagues²⁷ surveyed 150 caregivers of children with asthma in Baltimore to conduced spatial analysis identifying homes within 500 ft of a homicide to validate other subjective measures of neighborhood violence.

While the work of Curry and colleagues²⁶ and Tonorezos and colleagues²⁷ advanced an understanding of how objective and subjective measures of violence and more importantly perceptions of neighborhood disorder may influence psychological health, neither study accounted for potential protective characteristics of neighborhoods.^28,29 There is evidence that neighborhood and individual supportive mechanisms can have both direct effects and moderating effects for depression.^30,31 A study of 343 neighborhood clusters in 2 census tracts in Chicago found that most social support variables (residential stability, family structure, social cohesion, reciprocal exchange, and social ties) were associated with lower levels of depressive symptoms.³² In another study, perceiving one’s neighborhood to be high in collective efficacy, defined as trusting cohesive relationships and informal social control in a neighborhood, predicted healthier mental health trajectories independent of the presence of neighborhood problems.¹⁸ An earlier study by Shalowitz and colleagues³³ found that individual self-reported social support significantly contributed to maternal symptoms of depression among mothers of children with asthma.

In summary, understanding asthma morbidity in children is complex and is often associated with the context in which they live. Links have been found between maternal depressive symptoms, neighborhood violence, social support, and asthma outcomes. The intent of this study was to further elucidate the relationships between caregiver depressive symptoms and neighborhood subjective and objective measures of crime victimization, as well as perceptions of neighborhood satisfaction and sense of community in caregivers of children with poorly controlled asthma from predominantly low-income, racial and ethnic minority families. The overall objectives were as follows:

1. To examine the association between neighborhood violent crime victimization (objective and subjective measures) and depressive symptoms among caregivers of children with uncontrolled asthma.

2. To examine the association of neighborhood perceptions (satisfaction; a sense of community) and depressive symptoms in caregivers of children with uncontrolled asthma.

3. To examine whether neighborhood violent crime victimization and neighborhood perceptions interact and potentially depend on demographic and psychosocial characteristics (social support, life stress) in the association of each factor with caregiver depressive symptoms.

METHODS

This is a secondary data analysis of baseline data obtained from a randomized controlled trial (RCT) of a home-based environmental control intervention for children aged 3 to 12 years and diagnosed with persistent, uncontrolled asthma.³⁴ Families of children attending 2 large urban pediatric EDs for acute asthma exacerbations were recruited and enrolled during an acute asthma ED visit from August 2013 to February 2016. Inclusion criteria included physician diagnosed persistent, uncontrolled asthma based on national asthma guidelines³⁵ including caregiver report of asthma symptom frequency, having 2 or more asthma ED visits or at least 1 hospitalization over the past 12 months, and residing in the Baltimore metropolitan area. Children were excluded if they had a significant nonasthma respiratory condition. The RCT was approved by the 2 large institution institutional review boards and is registered with ClinicalTrials.gov (NCT01981564). Written informed consent was obtained from each child’s primary caregiver, and all children older than 8 years provided verbal assent to participate. Caregivers were interviewed by a research assistant blinded to group assignment at enrollment into the study using a survey that included health, sociodemographic, caregiver depressive symptom scores, environmental exposures, and neighborhood characteristics.

Measures

Individual correlates: Social support

Social support was assessed using the Medical Outcomes Social Support Survey (MOS-SS) consisting of 19 items.³⁶ For each item, respondents were asked about the availability of support (eg, “Someone to help you if you were confined to bed”; “Someone who hugs you”; and “Someone who understands your problems”), with 5 choices ranging from “none of the time” to “all of the time.” In this study, the Cronbach α was 0.97, which was previously demonstrated elsewhere.³⁴ In the current analyses, social support was dichotomized similar to previous studies, where poor support was defined as having an MOS-SS score less than the 25th percentile compared with those whose social support was the 25th percentile or greater.^37,38

Life stress

A visual analog scale (VAS) was used to ascertain the caregiver’s perception of general life stress over the past month. Moderate to high associations have been reported with VAS scores in the assessment of pain (r = 0.88–0.98).³⁹ Caregivers were asked to select a number between 1 and 10 to rate their perception of current life stress, with 1 indicating no stress and 10 reflecting the highest possible level of stress.

Caregiver depressive symptoms

Caregiver level of depressive symptomatology was measured using the 20-item Center for Epidemiologic Studies-Depression (CES-D) scale. The scale has high validity and good internal consistency ranging from 0.85 to 0.90.⁴⁰ The overall total CES-D score is a sum of all 20 items, ranging from 0 to 60. Participants reported their experience for each symptom in the last 2 weeks on a 4-point Likert scale (0 = not all, and up to 3 = every day or almost every day). This variable was treated as a dichotomous variable, with a cutoff score of 16 or greater indicating risk for depression.⁴⁰ The Cronbach α for the CES-D in this study was 0.90.

Neighborhood variables

Perceived Neighborhood Scale.

The Perceived Neighborhood Scale (PNS) was used to measure parental perceptions of neighborhood characteristics and has been shown to have high convergent and external validity.⁴¹ Three PNS subscales, (1) Satisfaction with Neighborhood, (2) Sense of Community, and (3) Perception of Crime, were administered to study participants. All subscales use a Likert response format (1 = strongly agree, 2 = agree, 3 = unsure, 4 = disagree, and 5 = strongly disagree), with lower scores reflecting more positive perceptions of neighborhood characteristics.

The Satisfaction with Neighborhood subscale (9 items) explores perceptions of how good the neighborhood was for raising a child (eg, “My neighborhood is a good place to live”; “My neighborhood is a good place to raise a child”). The Cronbach α was 0.84. The subscale was summed, with lower scores indicating higher satisfaction.

The Sense of Community subscale (7 items) pertains to perceptions about neighbors, feelings of belonging, and a sense of support (eg, “There are people I can rely on among my neighbors”; “People trust each other in my neighborhood”). The Cronbach α for the Sense of Community subscale was 0.87. The subscale was summed, with lower scores indicating a greater sense of community.

The Perception of Crime subscale (9 items) focuses on perceptions of social disorder including drinking and drug use in the neighborhood as well as perceptions of fear related to being raped, murdered, robbed, or mugged in the neighborhood (eg, “There is public drinking in my neighborhood”; “People are scared of being raped in my neighborhood”). Scores were summed, with high scores representing perceptions of increased crime since the individual items were reversed coded. The Cronbach α for the Perception of Crime subscale was 0.93.

Objective crime data: Violent crime victimization rate.

Our objective measures of neighborhood crime data were obtained by geocoding the child’s home address to census tracts within Baltimore City using the geospatial referencing software, ArcGIS version 10.5.1.⁴² Publicly available victim-based crime data were obtained from the Baltimore City Police Department (https://data.baltimorecity.gov/). These data were downloaded from 2013 to 2015, the same period that the objective data were collected, and were stratified to include violent crime victimizations (per the Federal Bureau of Investigation’s Uniform Crime Reporting Program) comprising homicide/manslaughter, rape, aggravated assault, and robbery.⁴³ Similar to previous studies, violent crime victimization rates were calculated per year by dividing the total number of violent crime victimizations in a census tract by the tract population and multiplying by 1000.⁴⁴

Data analysis

CES-D scores were dichotomized using the established cutoff score of 16 or higher to classify caregivers with high versus low/no depressive symptoms.⁴⁰ Baseline differences in categorical sociodemographic variables by caregiver CES-D scores (below cut point <16 vs higher than cut point ≥16) were tested for statistical significance by means of a chi-square test or Fisher’s exact test. Student t tests and Mann-Whitney U tests were used to examine the bivariate relationships of caregiver depressive symptoms and life stress, length of time in neighborhood, homicide rate, and neighborhood subscale variables including perception of crime, satisfaction with neighborhood, and sense of community.

Multiple logistic regression analysis determined the combination of factors that correspond to an increased likelihood of depressive symptoms. Factors were entered in order of significance in bivariate analyses, with consideration of all 2-way interaction effects and terms was retained in the final model when P < .05. All data analyses were performed using SPSS Statistical Software 22.⁴⁵

RESULTS

Baseline child and caregiver characteristics

Overall, 554 children and their caregivers were screened for study enrollment (Figure 1). Of these, 437 were eligible, resulting in 222 enrolled children and their caregivers who agreed to participate. In examining the caregivers who agreed to participate and those who did not, there were no significant differences by child age, neighborhood zip codes, or race/ethnicity between participating caregivers.³⁴ Finally, this analysis was limited to the families residing only in Baltimore City (n = 196; 88.3%) due to objective crime data only available for city residents.

Figure 1.

Recruitment diagram.

The caregivers’ mean age was 31.4 years (SD = 7.7), ranging from 18 to 62 years. Most caregivers were African American (94.9%), single (86.8%), female (97.4%), and had a high school education or more (80.5%), with 19.5% reporting less than high school education and 47.2% unemployed. Income was reported by category, with 29.5% of the caregivers reporting less than $10 000 per year and only 13.0% reporting $40 000 or greater and using the federal poverty index based on family income and number of family members⁴⁶; more than half (59.6%) of the families were below the poverty line. Most children were male (64.9%), African American (94.9%), and Medicaid health insured (94.9%), with a mean age of 6.4 years (SD = 2.7). The mean number of children per household was 2.5 (SD = 1.4),and the total mean number of household members was 4.6 (SD = 1.8).

Life stress/social support, depressive symptoms

Mean life stress score was 6.1 (SD = 2.9) based on a scale of 1 (low) to 10 (high). Social support was perceived to be very high with a median score of 91, with the range of possible scores of 0 to 100. The mean depressive symptom score was 12.6 (SD = 11.1). When the scores were dichotomized (<16 vs ≥16 total score), one-third (32.7%) scored 16 or greater on the CES-D scale, indicating high depressive symptoms (Table 1).

Table 1.

Demographic and Psychosocial Characteristics by Depressive Symptoms

Caregiver	Overall N=196	CES-D<16 N=132 (67.3%)	CES-D≥16 N=64 (32.7%)	Test Statistic P Value
Age Mean (SD)	31.4 (7.7)	31.9 (8.0)	30.4 (6.9)	t(186) =1.19, p=.234
Female %	97.4%	96.2%	100.0%	Fisher’s exact, p=.175
Married %	13.2%	16.0%	7.8%	X2(1)=2.47 p=.116
Poor (<Federal Poverty Level) %	59.6%	54.4%	70.2%	X2(1)=3.94 P=.047
Children who live in house Mean (SD)	2.5 (1.5)	2.6 (1.5)	2.4 (1.4)	t(194) =.73, p=.465
Education				X2(2)=11.70 P=.003
Some High school or less	19.5%	13.0%	32.8%
High school grad	41.5%	42.7%	39.1%
Some college or More	39.0%	44.3%	28.1%
Life Stress (mean)(SD)	6.1 (2.9)	5.2 (2.7)	7.8 (2.5)	t(194) = −6.25 p<.001
Social support	91.0	91.0	81.0	Z= −2.36, p=.018
Median (IQR)	(71.0, 95.0)	(76.0, 95.0)	(57.0, 95.0)

Note: CES-D = Center for Epidemiologic Studies-Depression; SD = standard deviation; IQR = interquartile range [25^th, 75^th percentiles].

Neighborhood characteristics

Neighborhood perceptions of satisfaction and a sense of community were assessed in several ways, with lower scores reflecting more positive perceptions on a Likert scale from 1 to 5. The scale for neighborhood safety was flipped in that on a scale of 1 to 5, higher scores reflected more perceived crime. The overall mean for the Satisfaction with Neighborhood subscale was 2.6 (SD = 0.8), and the overall mean for the Sense of Community subscale was 3.2 (SD = 1.1). The overall mean of the Perception of Crime subscale was 2.6 (SD = 1.0). At least one-third of the study participants agreed or strongly agreed that their neighborhoods had troublemakers hanging in their neighborhood (44.3%) or witnessed public drinking (34.3%). The safety questions asked whether friends and relatives felt unsafe (12.4% agreed) and whether people in their neighborhood were scared of people being robbed (26.7%), raped (15.9%), or being mugged (23.9%). More than one-third (38.4%) of the participants were unsure or disagreed that it was safe for their child to play outside. From an objective violence data perspective, there were an average of 20.2 (SD = 9.6) violent crime victimizations per 1000 people in a census tract.

Correlation analyses

To examine the bivariate relationships between objective violent crime and subjective perceptions of crime as well as neighborhood cohesion and satisfaction, we report Spearman’s correlations. The violent crime rate had a moderate but significant direct association with perceptions of crime (r_s = 0.36, P < .001).Perceptions of increased crime were positively but moderately related to a decreased sense of community (r_s = 0.40, P < .001) and positively and strongly related to decreased satisfaction with neighborhood (r_s = 0.75, P < .001). The objective measure of violence (victim rate) was moderately associated with lower satisfaction with neighborhood (r_s = 0.39, P < .001) and positively but weakly related with a decreased sense of community (r_s = 0.15, P < .044).

Bivariate relationships with depressive symptoms (unadjusted)

Only 2 demographic characteristics were related to depressive symptoms (Table 1). Poverty was significantly related to caregiver self-reported depressive symptoms (P = .047). Among caregivers with clinically significant depressive symptoms, 70% were characterized as being in poverty based on the federal poverty level.⁴⁶ Caregivers with high depressive symptoms were also more likely to report low education (some high school or less) than those with low depressive symptoms (32.8% vs 13.0%, P = .003). Life stress was also related to depressive symptoms (P < .001) as was social support (P = .018). Those with higher depressive scores had higher mean life stress and lower social support. Caregivers with depressive symptoms were less satisfied with their neighborhoods (P < .001) and reported a lower sense of community (P = .016) (Table 2). In looking at the objective data related to violent crime and caregiver depressive symptoms (Figure 2), the violent crime victimization rate corresponded directly to depressive symptoms (P = .005) (Table 2).

Table 2.

Caregiver Depression and Mean Values of Neighborhood violence and Neighborhood Perceptions

Caregiver	Overall N=196	CES-D<16 N=132 (67.3%)	CES-D≥16 N=64 (32.7%)	Test Statistic P Value
Satisfaction with neighborhood Mean (SD)	2.6 (0.8)	2.5(0.8)	3.0 (0.8)	t(191) = −3.68,P<.001
Sense of community Mean (SD)	3.2 (1.1)	3.1(1.0)	3.5 (1.1)	t(192) =−2.44,P=.016
Perception of crime Mean (SD)	2.6 (1.0)	2.3 (1.0)	3.0 (1.1)	t(191) =−4.06,P<.001
Victim rate per 1000 tract population Mean (SD)	20.2 (9.6)	18.9 (9.3)	23.0 (9.7)	t(194) =−2.86,P=.005

Note: CES-D = Center for Epidemiologic Studies-Depression; SD = standard deviation.

Figure 2.

Violent crime rate and caregiver depressive symptoms in Baltimore, MD, 2013.

Multivariate model

All variables with theoretical or bivariate significance were initially tested, but only the variables contributing to an increased odds of caregiver depressive symptoms (either main or interaction effects) are included in the final multivariate model (Table 3). The final model excluded variables that did not show a significant confounding or modifying effect such as age, gender, marital status, social support, and the community-level variables. Variables significant in the final model included a binary term for education (some high school or less vs more) and continuous factors: number of children in the household, life stress, perception of crime in neighborhood, and victim rate per 1000 residents. The model also included 2 interaction terms to reflect effect modification of life stress on the relationship between perception of crime in neighborhood and victim rate per 1000 residents on odds of depression.

TABLE 3.

Final Multivariate Logistic Regression Model describes the Combination of Factors Related to Odds of Depression (CESD≥16).

	Odds ratio (95% CI)
Variable	Final Model
Education (some High School or less vs. more)	7.04 (2.48, 40.0), p=.001
Number of Children in Household (per one unit increase)	0.69 (.53, .92), p=.010
Life Stress a	1.82 (1.38, 2.39), p<.001
Perception of Crime in Neighborhood (per one unit increase)
when Caregiver Reports:
Low Life Stress (value=2)	6.05 (1.85, 19.77), p=.003
Average Life Stress (value=6.1)	2.17 (1.30, 3.64), p=.003
High Life Stress (value=8)	1.35 (.92, 1.98), p=.121
Victim Rate per 1000 (per one unit increase)
when Caregiver Reports:
Low Life Stress (value=2)	1.17 (1.04, 1.32), p=.008
Average Life Stress (value=6.1)	1.07 (1.01, 1.13), p=.014
High Life Stress (value=8)	1.03 (.98, 1.07), p=.253

^aOdds ratio evaluated at mean value for objective and subjective measures of violence.

Note: Low life stress indicates a rating of 2 out of 10, average life stress indicates a rating of 6.1 out of 10, and high life stress indicates a rating of 8 out of 10, where 0 is no stress and 10 is the highest possible level of stress.

The final multivariate model revealed that caregivers with low education (some or less than high school) were 7.04 times (95% confidence interval [CI], 2.60–19.07) more likely to have depressive symptoms, while each additional child in the household reduced the odds by 31% (adjusted odds ratio [aOR] = 0.69; 95% CI, 0.53–0.92) (Table 3). Contrary to expectations, the community-level factors, sense of community and community satisfaction, as well as social support were not significant after adjustment for contributory factors and were not included in the final model. The strength of the relationship of increased violent crime rate and perceived crime in neighborhood to depression depended on the caregiver’s life stress, as shown in Figure 3. Each unit increase in life stress for average perceived crime and victim rate per 1000 residents observed in our study population increased the odds of depressive symptoms by 1.82 (95% CI, 1.38–2.39). Lower levels of life stress were associated with an increased effect of higher neighborhood crime rate and perceived crime on a higher likelihood of depression. When life stress was relatively low (value = 2), each unit increase in the victim rate per 1000 residents was associated with 17% increased odds of depression (aOR = 1.17; 95% CI, 1.04–1.32) but with decreased effect as caregiver’s life stress increased to high levels (values ≥8 on a scale from 1 to 10, P ≥ .05). Correspondingly, each unit increase in perceived crime in the neighborhood translated to 6.05 times higher odds of depression when life stress was low (95% CI, 1.85–19.77),with less effect with increasing life stress (P ≥ .05 when life stress ≥8).

Figure 3.

Describes interaction of life stress with perceptions of crime in neighborhood and victim rate per 1000 on probability of depression in the final multivariate logistic regression model.

DISCUSSION

This study extends an understanding of the differential associations among subjective and objective measures of violent crime victimization and caregiver depressive symptoms and uniquely contributes to the literature by identifying the role of life stress as a moderator of this relationship. The cumulative measure of victimization rate based on census tract crime data was significantly related to depressive symptoms for those individuals experiencing low to moderate levels of life stress. This was unlike the findings from previous studies examining objective measures of violence. Curry and colleagues²⁶ did not find a direct path between neighborhood level of violence and depressive symptoms, but violence was associated indirectly with psychological distress through 2 pathways: perceptions of neighborhood disorder and experiences of violence.²⁶ Tonorezos and colleagues²⁷ did not find that nearby homicide was significant in the final model predicting depression but fear of violence increased the odds of depression by 6.7 and actual victimization showed a possible trend toward depression. Subjective perceptions of neighborhood crime in this study corresponded to an increased likelihood of caregiver depressive symptoms only when life stress was low to moderate.

Measurement limitations may be one explanation for this differential relationship of perceived stress to depressive symptoms in this study. The single-item, self-report life stress measure used in this study reflects a perceived cumulative expression of stress and does not capture the specific nature of the stressors (ie, chronic or acute). For example, Grote and colleagues⁴⁷ found that the severity of chronic stressors accounted for more of the variance of depressive symptoms than acute stressors. Furthermore, type of the stressor (ie, housing instability, food insecurity, neighborhood violence, structural racism, discrimination) may be more important than a cumulative measure. Discrimination related to racism, which may or may not be overtly perceived as stress, has been found to relate to depressive symptoms.^48–50 Previous research has also examined the meaning and sources of stress for caregivers of children with asthma and found it to be multidimensional in nature, with poverty, exposure to community violence, and substandard housing contributing to the experience of stress and heightening risk for depression.^51–54 In the case of caregivers of children with asthma, unstable and poor housing stock may have particular relevance in the care of their children, with significant deleterious effects on asthma management and outcomes.^51,52 With high levels of perceived stress in this study, neighborhood violence does not appear to be as strongly associated with depressive symptoms, perhaps because of these competing priorities and/or specific stressors.⁵⁵ Another plausible explanation has been described as becoming emotionally numb to the environment and less reactive to increased chronic stressors.⁵⁶

The specific coping mechanisms used to address the stressors of living within a violent neighborhood, or the diminishment of adaptive coping mechanisms in addressing the stressors, may be alternative explanations. A recent study, for example, found that the concept “meaning in life” promoted a positive effect in reducing depressive symptoms and promoting a positive affect over time regardless of perceived stress. Meaning in life was assessed using a 14-item instrument examining beliefs and reflections about life.⁵⁷ Our study illustrates the complexity in understanding predictors of depressive symptoms. Although it was beyond the scope of this study to examine specific coping styles, it is clear that multiple factors are associated with depressive symptoms. There is a need, particularly within the African American population, to further examine what factors may mitigate the impact of violence in communities.⁵⁷

Although neighborhood characteristics, such as collective efficacy and social cohesion, have been proposed as protective mechanisms for diverse populations,^28,30,31 in our sample of caregivers of children with poorly controlled asthma, these neighborhood variables were not directly related to depressive symptomatology in the final model. Previous research has shown a moderating effect of positive neighborhood characteristics on the effects of violence^28,53,54,58; however, these relationships became nonsignificant after adjusting for other contributing factors in this study. Similarly, although a significant bivariate association was observed between social support and caregiver depressive symptoms, self-reported social support failed to predict depressive symptoms in the final model. Perhaps, a social support measure that asks about support in managing a child’s asthma would yield different results. In another study of neighborhood quality, stress, and social support among African American women, social support did not mediate a relationship between neighborhood quality and depressive symptoms but was instead related to perceived stress.⁵⁹ A previous study of an urban neighborhood with predominantly African American residents examined the relationship between household income and depressive symptoms. Complex relationships in this study were noted, with social support mediating the relationships between household income and depression, but stressors associated with neighborhood disorder and discrimination influenced depression directly regardless of income.⁶⁰

We also observed lower education level and fewer children in households to have direct associations with caregiver depressive symptoms in our final model. Other studies exploring correlates of depression likewise identified lower educational attainment as a risk factor.⁶¹ However, the relationship of family size to caregiver depressive symptoms has often been inconsistent. Sperlich and colleagues,⁶² for example, did not find an association between number of children and maternal depressive symptoms, while others have found an association with a larger number of children. In our study, an increased number of children in the family were not associated with depressive symptoms. One possible explanation might be that having a child with a chronic illness is more stressful than having an increased number of children.

Finally, in this sample of caregivers of children with uncontrolled asthma, a third (33%) endorsed clinically significant depressive symptoms. This rate is consistent with prior studies that examined the prevalence of depressive symptoms (27%–47%) among caregivers of children with asthma.^20,61,63,64 An earlier study using the same instrument to measure depressive symptoms in a similar sample of mothers of children with asthma in Baltimore, however, found a lower prevalence of 25% with depressive symptoms.²⁰

Limitations of study

This study is not without limitations. Our visual analog measure of life stress was an indicator of perceptions of overall stress and did not offer specific information on the type of stress being experienced (ie, acute or chronic), nor the source of stress. In addition, although our sample primarily comprised biological mothers (92%), this may limit our understanding of depressive symptoms in other caregivers such as fathers and grandparents. However, the results contribute to the understanding of social and biological stressors of an underserved community largely comprising single mothers. The sample was primarily caregivers of children with poorly controlled asthma from low-income, African American families, living in largely segregated neighborhoods, so the observed relationships in this study may not be generalizable beyond this population. This population experiences a disproportionately higher burden from increased asthma morbidity and mortality and therefore we felt it was important to focus on this group.

Finally, the cross-sectional nature of the data precludes making any conclusions about causality in the observed relationships. Direct interpretation of odds ratios (ORs) as a measure of expected increased odds of depression for one group relative to another should be viewed with caution considering the 95% CI around the OR estimate; this is particularly relevant when the interval is relatively wide, which could indicate one group had a small number of patients, although significant differences between groups were still detected at the .05 level.

CONCLUSION: IMPLICATIONS FOR PRACTICE

Our findings suggest that clinicians must recognize and understand that the “neighborhood comes through the door” when a family enters a clinical setting.⁶⁵ Our findings underscore the critical need to screen for community violence exposure. Furthermore, objective and subjective perceptions of violence should be considered in clinical encounters with this high-risk population. Taking a trauma-informed approach to care for the children and families exposed to violence is one means of addressing what may be seen as a mental health problem that would otherwise be ignored. This should include increasing provider awareness of the extent of violence occurring in communities, providing basic screening tools to assess the extent of this exposure and to then establish partnerships with providers who can provide appropriate support.⁶⁶

There is a need to assess for depressive symptoms as well as sources of life stress in caregivers of children with poorly controlled asthma. Addressing social determinants of health is an approach that has gained increasing attention and focus in primary care.^67,68 If stressors are identified, interventions can be tailored to address specific stressors when possible, providing levels of support as needed within the context of the environment.⁶⁹ This support may include referrals to mental health providers, community health workers, or social workers. It may include referrals for housing support as needed. It is also imperative to partner with families and communities to identify protective factors that ameliorate the potentially harmful effects of neighborhoods and provide resources for stress reduction and mental health support, particularly for caregivers of children with a chronic illness. This might include partnering with local health departments that offer asthma home visits.⁷⁰ It may include partnerships as described by the Centers for Disease Control and Prevention to develop a vision, build awareness, and develop strategies among stakeholders to achieve safe communities.⁷¹ Ultimately, a systems-level approach, tackling the social determinants of health, advocating for violence prevention in all neighborhoods, and improved neighborhood conditions, will undoubtedly be required to address the larger issues critical in addressing asthma disparities within our communities.