The Neighborhood Environment and Hispanic/Latino Health

Abstract

Purpose:

Hispanic/Latino adults on the Texas-Mexico border have high rates of chronic disease. Neighborhoods can influence health, though there is a limited research on neighborhood environment and health in Hispanics/Latinos. The purpose of this study was to assess the relation of neighborhood environment with health variables in Hispanic/Latino adults, including physical activity [PA], depression, anxiety, and lab-assessed conditions (type 2 diabetes, metabolic syndrome, and chronic inflammation).

Methods:

Participants were randomly-selected from a Hispanic/Latino cohort on the Texas-Mexico border. Neighborhood environment, self-reported PA, anxiety, and depression were assessed through questionnaires. Laboratory values determined Type 2 diabetes, metabolic syndrome, and C-reactive protein (CRP). We conducted multivariable linear and logistic regression analyses to assess the associations of neighborhood environment and health variables, controlling for covariates.

Results:

Participants (n = 495) were mostly females, without insurance. After controlling for covariates, crime (Adjusted Odds Ratio [AOR] = 1.59 (95%CI 1.06–2.38), no streetlights (AOR = 1.65, 95%CI 1.06–2.57), and traffic (AOR = 1.74, 95%CI 1.16–2.62) were all significantly associated with anxiety. Only traffic was significantly associated with depression (AOR = 1.61, 95%CI1.05–2.47). A lack of nearby shops (AOR = 0.57, 95%CI 0.38–0.84) and no one out doing PA (AOR = 0.53,95% CI 0.34–0.83) were both significantly associated with lower odds of meeting PA guidelines. A lack of nearby shops was associated with a 26% increase in the CRP value (β = 0.26, 95%CI 0.04–0.47).

Discussion:

Several neighborhood environment variables were significantly associated with mental health, PA and CRP, though estimates were small. The neighborhood environment is a meaningful contextual variable to consider for health-related interventions in Hispanic/Latino adults, though more study is needed regarding the magnitude of the estimates.

Trial registration:

NCT01168765.

Purpose

Hispanic/Latino (heretofore Latino) adults, especially those living on the Texas-Mexico border, have high rates of obesity, type 2 diabetes, and metabolic syndrome compared to non-Hispanic whites.^1–3 While the causes for obesity and obesity-related conditions are multifaceted and involve the influence of the individual’s knowledge and beliefs, along with the healthcare access and social environment, neighborhoods can also exert an influence on health.⁴

There is a limited body of research on the association of perceived neighborhood environment, such as traffic, stray dogs, and lack of appropriate walking infrastructure, with various health variables in Latinos. In the general population, the neighborhood environment is associated with physical activity, type 2 diabetes, metabolic syndrome, and various other health variables.⁵ The bulk of the existing research in Latinos has focused on the association of the built environment with physical activity, and that association is mixed.^6–10 The neighborhood environment can also impact mental health, including depression and anxiety.¹¹ More specifically, negative aspects of the neighborhood environment may lead to individual psychological stress, including depression. Few studies have been conducted in Latino adults on the association of neighborhood environment with more physiological outcomes, including metabolic conditions. Of these, 2 studies found that the neighborhood environment was not associated with diabetes.^12,13 and was associated with metabolic syndrome in Latinos,¹⁴ but there is a need for additional research on how neighborhood environment is associated with a variety of health-related variables, including chronic inflammation, which contributes to many adverse health outcomes.

The purpose of this study was to assess the association of neighborhood environment with physical activity, mental health (depression and anxiety), metabolic conditions, including type 2 diabetes and metabolic syndrome, and chronic inflammation among Latinos. As an exploratory aim, we also assessed depression and anxiety as mediators in the relationship between neighborhood environment and other health variables.

Methods

Design

Data for these analyses come from a randomized control trial, Tu Salud ¡Si Cuenta! (Your Health Matters!) at Home Intervention.¹⁵ Participants in this study were randomly selected from the Cameron County Latino Cohort (CCHC). CCHC is a cohort of Latino individuals who were randomly selected and recruited from census tracts and blocks across socio-economic status (SES) levels from primarily 1 medium-sized community on the Texas-Mexico border.¹⁶ After random selection from this cohort, participants were randomized to either receive the intervention (n = 250) or standard care arm (n = 250). The intervention included 6 monthly community health worker (CHW) visits to the home, during which the CHW would deliver an educational and motivational intervention focused on physical activity and healthy food choice. The trial is registered with clinicaltrials.gov (NCT01168765).

Sample

Data from the current analyses comes from the baseline assessments of the 500 participants of the aforementioned randomized control trial.¹⁵ Baseline data was collected prior to the start of the intervention, so we collapsed both groups into a single sample (n = 500). The city from which this sample was drawn sits on the U.S./Mexico border and is the county hub. It is a city of about 180,000 and generally is considered one of the poorest communities in the United States.¹⁷ The neighborhood contexts for the sample range from older downtown neighborhoods, to suburban neighborhoods with some amenities, to recently improved colonias within city limits that now have paved roads and some street lights, to homesteads on the out-skirts of the city that have livestock and larger lots. The violent crime rate in this city is low,¹⁸ people in the household is fairly high (3.47 per household compared to 2.62 in the U.S.) and the population density is also low (1,322.6 per square mile compared to 3,517.6 in Dallas, Texas), according to the Census 2015–2019 data.¹⁷

Measures

Data were collected by interviewer-administered questionnaires delivered by bilingual, trained staff at baseline.

Covariates.

Questionnaires included demographic questions, including age (18–29, 30–39, 40–49, and ≥50 years), sex, insurance (yes, no), language preference (English, Spanish, Bilingual), educational attainment (≤8 years, >8 years of school), marital status (married, other), and employment status (employed, unemployed). Acculturation was assessed with the 4-item language-use subscale of the Short Acculturation Scale for Latinos.¹⁹

Neighborhood environment: independent variables.

We assessed neighborhood environment with 14 items adapted from the Physical Activity Neighborhood Environment Scale,²⁰ which includes questions about crime, presence of sidewalks, traffic, and other characteristics about the neighborhood. Participants answered either yes or no to each item, and we scored items so that a 1 represented responses seen as negative aspects of the neighborhood environment or problems. When possible, we created constructs, which included Unfriendly for Pedestrians, Crime Safety, and Stray Dogs. Unfriendly for Pedestrians (Cronbach’s Alpha = 0.71), which included 3-items assessing the lack of crosswalks and pedestrian signals, lack of sidewalks on most streets, and lack of well-maintained sidewalks. Crime Safety (Cronbach’s Alpha = 0.68) included 3 items that assessed if there was high crime generally, and then if crime made it unsafe to walk during the day, or if it made it unsafe to walk at night. Stray Dogs (Pearson’s Correlation Coefficient = 0.73) included 2 items that assessed whether stray dogs made it difficult or unsafe to walk and if they made it difficult or unsafe to use the recreation facilities. Other items were retained as single-item constructs, as their internal consistency was unacceptable when grouped together. These items included questions about whether there were shops in easy walking distance of their home (Lack of Shops), if there was a bus stop within 15 minutes walking distance from their home (No Bus Stop), if the neighborhood had free or low cost recreation facilities (No Recreation Facilities), if the streets were well-lit at night (No Streetlights), if there was so much traffic on the streets that it made it difficult or unpleasant to walk (Traffic) or there were other people out being physically active in the neighborhood (No one out doing physical activity).

Mental health.

Anxiety was assessed with the Zung Self-Rating Anxiety Scale and scored used standardized procedures.²¹ The range of scores were 25–90, with higher scores indicating more anxiety symptoms. We assessed depressive symptoms with the Center for Epidemiologic Studies Depression Scale and it was scored using standardized procedures.²² The range of scores were 0–54, with higher scores indicating more depressive symptoms.

Physical activity.

Total minutes of leisure-time moderate-to-vigorous physical activity was assessed using either the International Physical Activity Questionnaire (IPAQ)²³ or a modified Godin Leisure-Time Exercise Questionnaire,²⁴ with both questionnaires collecting the frequency, intensity, and duration of physical activity. A small sample completed the IPAQ (n = 16) at the initiation of the trial before moving to the Godin measure that more effectively assesses physical activity interventions. Based on the MET minutes calculated from either instrument, we created a dichotomized variable based on meeting the 2018 Physical Activity Guidelines for Americans, or not,²⁵ and used this variable for analyses.

Metabolic health.

Anthropometric measures, including blood pressure and hip and waist circumference, as well as blood samples were collected by study staff. We took 2 separate supine blood pressure readings as well as 2 measures of hip and waist, with the average of the measurements used for analyses. Hips were measured at the widest part of the hips and waist was measured at the umbilicus. Ten-hour fasting blood samples were collected to examine total cholesterol, low density lipoprotein (LDL), high density lipoprotein (HDL), HbA1c levels, triglyceride levels, insulin, and fasting glucose. Blood specimens were processed by a commercial Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory.

Per the American Diabetes Association, participants were classified as diabetic if they had a Mean Fasting Plasma Glucose ≥126 mg/dL, HbA1c ≥6.5%, or reported taking medications for diabetes.²⁶ ATP III guidelines were used for metabolic syndrome criteria.²⁷ Men with waist circumference >40 inches and women with waist circumference >35 inches were categorized as having abdominal obesity. Participants with triglycerides ≥150 mg/dL were categorized as having elevated triglycerides. Participants with HDL <40 mg/dL in men and <50mg/dL in women were categorized as having low HDL. Those with ≥130 mm Hg systolic or ≥85 mm Hg diastolic blood pressure or taking antihypertensive medication were categorized as having elevated blood pressure. And lastly, those with Mean Fasting Plasma Glucose ≥100 mg/dL or taking hypoglycemic medication were categorized as having elevated fasting glucose. If participants met any 3 of these 5 criteria, they were categorized as having metabolic syndrome.²⁷

Inflammation.

Ten-hour fasting blood samples were collected to examine high-sensitivity C-reactive protein (CRP). Blood specimens for CRP were processed at an academic field research laboratory using enzyme-linked immunoassay (ELISA) with external quality control performed on randomly selected samples at another academic laboratory. High levels of high-sensitivity CRP was defined as >3 mg/L, which is indicative of chronic inflammation and a high risk of cardiovascular disease.²⁸ We also excluded values >10 mg/L (n = 66), as these tend to be an indicator of acute inflammation, rather than chronic inflammation. These data were only collected on a sub-set of the participants. For analyses with CRP, we performed a log transformation as the data were skewed.

Analysis

Univariable associations were assessed between all neighborhood problems, with physical activity, mental health, inflammation and metabolic health outcomes. We performed a series of multivariable linear (log-transformed CRP) and logistic (all other variables) regression to assess the associations of neighborhood environment and health outcomes, controlling for potential confounding effects based on the existing literature in this area,^8,12,29 such as gender, age, marital status, education, and insurance, that have been known. We tested the parameter estimate of the variance at the census block level, and this parameter was small and not statistically significant, so it was not included in analyses. Additionally, we used Structural Equation Modeling (SEM) to conduct path analyses to examine specific relationships between the variables of interest.^30,31 For these analyses, we used the SAS Covariance Analysis of Linear Structural Equations (CALIS) procedure,³² in which we hypothesized functional or causal relationships among the observed variables, specifically depressive symptoms and anxiety as mediators of the association of neighborhood environment with metabolic health outcomes, CRP and physical activity. Based on the findings from the path analysis, we generated path diagrams to visualize the models and statistical results, and reported parameter estimates of the direct, indirect and total effect for the variables. We assessed whether anxiety or depression is a mediator of the association of interest by comparing the magnitude and significance of the indirect effect with those of the direct effect. Participants with missing dependent variables (25% of CRP variable because it was only collected on a subset of participants and we excluded values >10 mg/L; <5% all other variables) were excluded in the multivariable analyses. Statistical significance was set at p < 0.05. All analyses were performed on SAS^® v. 9.4 (SAS Institute Inc., Cary, NC).

Results

Participants (n = 500) were mostly female (70%), without insurance (68%) and had only completed 8th grade (69%) (Table 1). Most participants were under the age of 49 (74%) and preferred Spanish to English (76%). Almost half (49%) were employed. About 28% of the sample experienced depressive symptoms and 32% had symptoms of generalized anxiety disorder. Just over 35% of the sample was meeting physical activity guidelines. Based on lab assessments, 36% of the sample had metabolic syndrome, with an average of 2 of the 5 metabolic syndrome characteristics across the entire sample, and 23% of the sample had type 2 diabetes.

Table 1.

Sample Characteristics.

Variable n (%)	n	%
Gender
Male	151	30.20%
Female	349	69.80%
Age, years
18–29	111	22.20%
30–39	141	28.20%
40–49	116	23.20%
≥50	132	26.40%
Uninsured	341	68.20%
Language
Spanish	376	75.50%
English	58	11.65%
Bilingual	64	12.85%
Education	345	69.00%
≤8	155	31.00%
>8	345	69.00%
Married	328	65.60%
Employed	243	48.60%
Depression	136	27.98%
Anxiety	157	32.44%
Elevated C-Reactive Protein*	205	65.67%
Diabetes	111	22.89%
Metabolic Syndrome	181	36.20%
Number of Metabolic Syndrome Criteria, Mean (SD)	2.07	1.09
Meeting physical activity guidelines	173	35.09%
Unfriendly for pedestrians	302	60.40%
Crime	188	37.60%
Stray Dogs	176	35.20%
Lack of Shops	236	47.58%
No Bus Stop	94	19.22%
No Recreation Facilities	168	34.08%
No Streetlights	132	27.05%
Traffic	312	36.84%
No one out doing PA	148	30.08%

Note:

^*We excluded observations with values >10 mg/L.

Most of the sample reported that their neighborhood was unfriendly for pedestrians (60%) and almost half reported that there were very few shops in easy walking distance of their home (48%).Only around 1/3 of the sample reported crime, stray dogs, no recreation facilities, no streetlights, traffic, or no one out doing physical activity as a problem.

Main Findings

At baseline, controlling for covariates, crime (Adjusted OR = 1.59, 95% CI 1.06–2.38), no streetlights Adjusted OR = 1.65, 95% CI 1.06–2.57), and traffic (Adjusted OR = 1.74, 95% CI 1.16–2.62) were all positively associated with anxiety (Table 2). Only traffic was significantly associated with depressive symptoms, with the odds of depressive symptoms 1.6 times higher in those reporting traffic as a neighborhood problem than those not (Table 2). A lack of nearby shops (Adjusted OR = 0.57, 95% CI 0.38–0.84) and no one out doing physical activity (Adjusted OR = 0.53, 95% CI 0.34–0.83) were both associated with lower odds of meeting PA guidelines. A lack of nearby shops was associated with a 26% increase in CRP (Adjusted b = 0.26, 95%CI 0.04–0.35). Unexpectedly, those reporting a lack of streetlights as a problem had 40% lower odds of metabolic syndrome compared to those not reporting this problem.

Table 2.

Multivariable Associations Between Neighborhood Stressors and Various Health Variables.

Variable	Anxiety		Depression		Meeting PA guidelines		Elevated C-reactive protein (Log)a		Type 2 diabetes		Metabolic syndrome
	Adjusted OR	CI	Adjusted OR	CI	Adjusted OR	CI	Adjusted mean difference	CI	Adjusted OR	CI	Adjusted OR	CI
Unfriendly for pedestrians	0.85	(0.57–1.28)	1.18	(0.77–1.82)	0.79	(0.53–1.18)	−0.01	(−0.24–0.21)	0.98	(0.60–1.58)	0.70	(0.48–1.03)
Crime	1.59*	(1.06–2.38)	1.21	(0.79–1.85)	0.76	(0.50–1.14)	−0.15	(−0.37–0.08)	1.02	(0.63–1.65)	1.19	(0.81–1.76)
Stray Dogs	1.11	(0.74–1.67)	1.12	(0.73–1.72)	1.09	(0.72–1.65)	−0.16	(−0.39–0.08)	0.89	(0.54–1.45)	0.95	(0.63–1.41)
Lack of Shops	0.79	(0.53–1.18)	0.86	(0.57–1.31)	0.57**	(0.38–0.84)	0.26*	(0.04–0.47)	0.92	(0.58–1.45)	1.15	(0.79–1.68)
No Bus Stop	1.09	(0.66–1.80)	1.44	(0.86–2.42)	0.95	(0.57–1.57)	0.24	(−0.05–0.51)	1.01	(0.57–1.78)	1.22	(0.76–1.96)
No Recreation Facilities	1.49	(0.98–2.27)	1.41	(0.91–2.19)	0.83	(0.54–1.27)	0.04	(−0.20–0.27)	1.00	(0.62–1.62)	0.92	(0.62–1.38)
No Streetlights	1.65*	(1.06–2.57)	1.25	(0.78–2.00)	0.76	(0.48–1.20)	0.07	(−0.18–0.32)	0.58	(0.33–1.02)	0.60*	(0.38–0.95)
Traffic	1.74*	(1.16–2.62)	1.61*	(1.05–2.47)	0.84	(0.56–1.27)	−0.05	(−0.27–0.18)	l.l1	(0.69–1.79)	1.18	(0.80–1.75)
No one out doing PA	0.94	(0.61–1.45)	1.12	(0.71–1.75)	0.53*	(0.34–0.83)	0.03	(−0.20–0.27)	0.72	(0.43–1.20)	0.86	(0.56–1.30)

Note: all analyses controlled for gender, age, marital status, education, and insurance;

^*p < 0.05,

^**p < 0.01.

^a66 Participants with CRP >10 were excluded for this particular analysis.

Neighborhood environment was not significantly associated with type 2 diabetes. Furthermore, presence of stray dogs, no bus stop within 15 minutes from home, and the neighborhood lacking pedestrian-friendly infrastructure were not significantly associated with physical activity, mental health, inflammation, or metabolic health outcomes.

Mediation

To identify mechanisms of the association between neighborhood environment and the various health outcomes, we also assessed depressive symptoms and anxiety as mediators. However, most all indirect path estimates were minimal (i.e., the indirect path estimates were smaller than those of direct paths), indicating that anxiety and depressive symptoms do not mediate the association of neighborhood environment with metabolic conditions, inflammation, and physical activity. A few models did show some small-to-moderate indirect associations. For example, as seen in Figure 1A, though the indirect path estimate did not reach significance, the indirect association between crime and metabolic syndrome was about 40% of total association (direct path estimate = 0.02, p = 0.659; indirect path estimate = 0.013, p = 0.079). A pattern can be seen with similar Figure 1B, where there appears to be some association between traffic and metabolic syndrome through anxiety (direct path estimate = 0.013, p = 0.768; indirect path estimate = 0.016, p = 0.055).

Figure 1.

Standardized regression coefficients for the relationship between neighborhood variables and metabolic syndrome, as mediated by anxiety. Notes: all analyses are adjusted for gender, age, language, marital status, education, and insurance; *p < 0.05, **p < 0.01.

Discussion

Our findings provide insight into an understudied population on specific features of the perceived neighborhood environment associated with health variables in Latino adults on the Texas-Mexico Border. In an under-resourced, medium-sized community with a primarily uninsured and female Latino sample, our results indicate that various aspects of the perceived neighborhood environment were associated with mental health and chronic inflammation. In this sample, crime, no streetlights, and traffic were all positively associated with anxiety and traffic was also associated with presence of depressive symptoms. These findings align with previous research that has found that a composite score of unfavorable neighborhood characteristics was positively associated with psychological distress, which includes depression.³³ While Chirinos and colleagues also found that neighborhood characteristics were associated with CRP,³³ our study found that only a lack of nearby shops was associated with CRP. Studies on the association of specific neighborhood environment with mental health and inflammation have been limited, and more research is needed with Latino samples.

Our study also provides insight on the specific neighborhood features in this community that were associated with physical activity. We did not find any significant association of sidewalks or perceived crime with physical activity. These findings contrast a previous study by Fields and colleagues using similar constructs; they found that among Latinos adults in Kansas, the presence of sidewalks was positively associated with meeting physical activity guidelines, while crime was inversely associated.⁶ A similar construct, the walking environment, was significantly associated with meeting physical activity guidelines among Latinos in Massachusetts.⁸ Silfee and colleagues also found that having shops nearby was associated with active commuting. While we did not assess domains of physical activity in this study and therefore cannot draw any conclusions related to active commuting, our findings mirror those of Silfee and colleagues as we did find a significant inverse association between a lack of nearby shops and meeting physical activity guidelines. Lastly, we found that seeing people participating in physical activity was an important variable associated with meeting physical activity guidelines, similar to the findings among Latinas in Chicago and North Carolina,^7,9 though this variable was not significantly associated with physical activity in Latinos adults in Kansas.⁶ Our findings add to the body of literature on the neighborhood environment’s influence on Latino physical activity, work which shows that the relevance of these various constructs is highly context-specific.³⁴ The neighborhood environment is nestled within the socio-ecological model, in which there are multiple levels of influence on individual-level behaviors and health outcomes. Though we did find significant associations, our estimates were small, indicating that neighborhood environment is relevant to the health of Latinos and is one level of influence on physical activity; however, it is not solely responsible for health outcomes, as physical activity is impacted by multiple levels of influence.³⁵ These findings align with other research in this area.⁵

We found that, as with other studies with Latinos, neighborhood environment was not associated with type 2 diabetes.^12,13 Unlike findings from Jospeh & Vega-López,¹⁴ in our study, we found neighborhood environment was not associated with metabolic syndrome, pointing to the potential for a minimal direct contribution of neighborhood environment to metabolic health. To attempt to understand mechanisms through which neighborhood environment may affect health outcomes indirectly, we also explored select mediators. While neither traffic nor crime were directly associated with metabolic syndrome, they were both indirectly associated through anxiety. By increasing an individual’s anxiety, neighborhood environment could be associated with metabolic syndrome in Latinos. It is also possible that there are other neighborhood characteristics that were not measured in this study that would show a stronger association to metabolic health. Thus, additional future research is needed in this area to explore causality and additional neighborhood characteristics, as well as more rigorously assess additional potential mediators to understand the mechanisms.

Strengths and Limitations

There were several limitations to this study. This was a cross-sectional study, which limited the interpretation of our results and precludes the ability to inferences about causality and temporal precedence. In this study, we only associated and tested self-reported moderate-to-vigorous intensity physical activity. There are known issues with the self-reporting of physical activity data, including overreporting. It is also possible that walking, rather than moderate-to-vigorous physical activity, may have been a more useful outcome that is directly related to the neighborhood environment.³⁶ This study only assessed perceived neighborhood environment, rather than the objective built environment. However, research suggests that is it the perceived, rather than the actual environment, that is more critical for health.³⁷ Lastly, participants came from different census tracts and blocks across Brownsville, Texas, which is a mix of downtown, suburbia and country suburbia, with large lots and farm animals. Thus, participants may not have all had the same type of environment and the nuances of their environment may not have been captured by our questions. However, we examined the parameter estimate of the variance at the census block level and found this estimate was both small and non-significant. Furthermore, in sensitivity analyses using mixed effect modeling to account for census block, results were very similar, with no substantial differences, to the results we have presented. This study’s strength lies in the examination of many important health variables and their interrelation in an understudied and underserved community.

Conclusions

We found that several perceived neighborhood environment variables were significantly associated with mental health, physical activity and CRP, and these estimates were small. We also found that some features of the neighborhood environment were not significantly associated with any of the mental or physical health outcomes measured. This study provides new insight into the nuanced role that neighborhood environment may play not only in a low-resource Latino community, but in other communities as well. Future research in a wide-range of communities, especially diverse, predominately minority, and/or low SES communities should consider the perceived neighborhood environment to begin examining patterns across and uniqueness within communities. The frontier of understanding the neighborhood environment and building frameworks for influence and causality related to health in diverse communities is uncharted.

So What?

What is already known on this topic?

We know that the neighborhood environment is associated with various health variables in Hispanic/Latino adults, however the data is mixed. While it has been associated with physical activity, there is less evidence for associations with mental health and metabolic conditions.

What does this article add?

The results of this study found that several specific neighborhood environment variables were associated with mental health, physical activity and chronic inflammation in Hispanic/Latino adults, while there was no association with metabolic conditions.

What are the implications for health promotion practice or research?

The neighborhood environment is a meaningful contextual variable to consider for health-related interventions in Hispanic/Latino adults, though more study is needed regarding the magnitude of the estimates.