Neighborhood Social Cohesion and Serious Psychological Distress Among Brazilian Immigrants in Boston

Abstract

Recent migrants to the United States face various stressors, including adjustment to new community norms and practices. To ease this transition, migrant groups have traditionally formed enclaves where they might live in close proximity and access institutions designed to serve their cultural interests. For newer migrant groups, such as Brazilians residing in New England, neighborhood social cohesion may therefore be particularly important for buffering against serious psychological distress. We use representative data from the 2007 Boston Metropolitan Immigrant Health and Legal Status Survey to estimate the association of serious psychological distress with neighborhood-level social cohesion among foreign-born Brazilian adults. We find that serious psychological distress is inversely related to neighborhood social cohesion (OR 0.66, 95% CI 0.46, 0.94). Annual earnings were also negatively associated with distress (OR 0.97, 95% CI 0.93, 0.99). Our findings suggest that neighborhood social ties may buffer against serious psychological distress for Brazilian migrants in New England.

Introduction

Mental disorders are widespread among foreign- and native-born residents of the United States (Carlat 2010; Gaines 1998; Vega and Alegría 2001; Watters 2010), and residential mobility (e.g., international migration, urbanization) is an important source of stress (Beard 1972 [1881]; Gatrell 2002; Marmot and Syme 1976; Rosenberg 1961). In high income countries, mental illness is the leading cause of disability, surpassing even cardiovascular disease and cancer (Reeves et al. 2011), and when combined with chronic disease, mental illness has been shown to accelerate morbidity and mortality (Moussavi et al. 2007). U.S. national-level surveys indicate that one in four adults suffer from a mental disorder each year and the lifetime prevalence of having had any disorder outlined in the Diagnostic and Statistical Manual of Disorders (DSM-V) is estimated to be 46% (Reeves et al. 2011).

However, these numbers likely understate distress in the population. While the DSM-V may capture clinical cases of discrete mental disorder, its symptom categories often fail to identify subclinical levels of distress from which most distress-related illnesses requiring medical treatment emanate (Dohrenwend et al. 1980; Rose 1992). Measures of nonspecific psychological distress, such as Kessler’s K6 scale of serious psychological distress (Kessler et al. 2002), therefore, have been developed to identify segments of the population exhibiting potentially unhealthy, albeit non-case, levels of stress. This may be especially important for discerning mental distress in populations, like Latin American migrants, that may more often present with somatic symptoms related to mental illness (López 2002; Vega and Alegria 2002). Psychological distress can further be understood as the consequence of social conditions and disparity in socio-environmental exposures (Mirowsky and Ross 2003b), therefore making it a useful measure for understanding mental health in relation to social interaction and neighborhood-level exposures.

Approximately 3.4% of the U.S. population (currently about 11 m Americans) reported being seriously psychologically distressed in 2017 (National Center for Health Statistics 2018b). Among foreign-born adults the prevalence rate was higher at 4% (National Center for Health Statistics 2018a), and among foreign-born Brazilian adults in the Boston Metro area, where the 2nd largest population of Brazilian migrants in the United States resides, the rate has been estimated at 7.3% (Marcelli et al. 2009a; U.S. Census Bureau 2017). In addition to generating social, emotional and economic disruptions in peoples’ lives (Mirowsky and Ross 2003c), psychological distress has been associated with numerous disease processes and outcomes, such as depression (Cairney et al. 2007), obesity (Zhao et al. 2009), cancer (Zabora et al. 2001), diabetes (Dharmalingam 2005), sleep quality (Hill et al. 2009), smoking (Cosci et al. 2009), cardiovascular disease, (Hamer et al. 2008; May et al. 2002; Nabi et al. 2008; Stansfeld et al. 2002) systemic inflammation (Goldman-Mellor et al. 2010; Holmes and Marcelli 2012; Puustinen et al. 2011) and multimorbidity (Fortin et al. 2006). The relatively high prevalence of serious psychological distress (SPD) observed among metropolitan Boston’s foreign-born Brazilian adults provides an opportunity to investigate whether the neighborhood in which one resides is associated with SPD.

Neighborhoods are environments in which factors that are thought to be associated with distress merge, including material circumstances, social interaction, household dynamics, health behaviors, and residents’ individual exogenous characteristics (Mirowsky and Ross 2003c; Sampson 2012). Sampson (2012) has argued that rather than people choosing their neighborhoods, “neighborhoods choose people…”; in other words, there are existing structural factors that influence neighborhood contexts over time, such as zoning laws or discriminatory housing practices, as well as constraints on the capacities of underserved populations to afford housing in certain neighborhoods. Additionally, how existing residents respond to newly arriving neighbors, especially those from other countries, may also affect mental health (Leu et al. 2011). The literature on residential segregation has long demonstrated a tendency on the part of some ethno-racial groups to relocate rather than remain in neighborhoods with changing demographic characteristics (Acevedo-Garcia and Lochner 2003; Denton and Massey 1991; Sampson 2013). Alternatively, for migrants, there may be a cultural draw to residing in neighborhoods that are home to their compatriots and where language barriers may be less and goods from home more available, easing the physical and mental transition (Logan et al. 2002; Massey 1990). All of these factors are likely to have an impact on the psychological health of neighborhood residents; and this in turn may exacerbate, instigate or buffer the occurrence of chronic disease (Lazzarino et al. 2013).

Yet there remain few studies examining potential relationships between neighborhood social cohesion—the trust between neighborhood residents and the extent to which they share values, norms and resources—and psychological distress among residents, and especially within specific subpopulations (Sampson et al. 1997). What research does exist suggests that social ties with neighbors and social cohesion impact the mental health outcomes of residents. For instance, those in neighborhoods with lower levels of social cohesion across a variety of metropolitan areas have been found to exhibit higher rates of depression (Echeverría et al. 2008) and mental disorder (Stockdale et al. 2007), and those experiencing higher levels of cohesion have lower levels of stress and distress (Cutrona et al. 2000; Holmes and Marcelli 2012, 2014; Kruger et al. 2007; Ross and Jang 2000).

In this paper we use the 2007 Harvard UMASS-Boston Metropolitan Immigrant Health and Legal Status Survey (BM-IHLSS) data to investigate how neighborhood social cohesion may influence SPD among legal and unauthorized adult Brazilian migrants residing in the Boston-Cambridge-Quincy Metropolitan Statistical Area (BCQ MSA). Specifically, we hypothesize that neighborhood social cohesion is negatively associated with SPD.

In our model we specified a number of covariates which have been shown in past research to be related to serious psychological distress, including socio-demographic characteristics and health behavior and status metrics. Regarding the former, we included age and sex, which are highly correlated not only with distress (Mirowsky and Ross 2003a), but with health outcomes in general, as well as skin color. Skin color is a more appropriate measure than race/ethnicity of phenotypic variation among Brazilian migrants as race is not assessed the same way in Brazil as in the United States, and while all of the migrants in our sample were the same ethnicity, there was a wide range of personally identified skin colors (Massey and Martin 2003; Telles 2002, 2004). Skin color has the potential to affect labor market and social outcomes, in particular, which in turn impact mental health (Rosenblum et al. 2015; Telles 2004).

We also adjusted for marital status, annual earnings and unauthorized legal status; being married has been shown to reduce the likelihood of distress as have higher earnings (Fiscella and Franks 2000; Mirowsky and Ross 2003c). These are also especially relevant factors for Brazilian migrants who largely came to the United States for work opportunities and sometimes had to leave their families behind to do so (Marcelli et al. 2009a). We are unaware of any other study evaluating the relationship of unauthorized legal status to psychological distress among Brazilian migrants, but given the tenuous nature of their residence in the United States and the fear and apprehension that generates, we hypothesized that such added stress may translate into serious psychological distress. Finally, serious psychological distress is a multivariate process that impacts, and is in turn impacted, by health behaviors and biological health conditions. In particular, obesity and poor sleep have been positively associated with distress (Burdette and Hill 2008; Friedman et al. 2002; Hill et al. 2009), while physical activity has been shown to mitigate distress (Hamer et al. 2008; Hamer et al. 2009). We therefore adjusted for body mass index, inadequate sleep patterns and moderate-to-vigorous physical activity in our models.

Methods

Study Population

The 2007 Harvard-UMASS Boston Metropolitan Immigrant Health and Legal Status Survey (BM-IHLSS) was a community-based biodemographic research study implemented in the Boston-Cambridge-Quincy, MA-NH Metropolitan Statistical Area (BCQ MSA) (Brown et al. 2005; Marcelli et al. 2009a, b; Minkler and Wallerstein 2003). The BM-IHLSS was the first random household survey to collect both legal status and biological data from any foreign-born population in the USA, and builds on the 1994 and 2001 Los Angeles County Mexican Immigrant Legal Status Surveys (LAC-MILSS I & II) pioneered by (Marcelli 2014); Marcelli and Heer (1997). Survey respondents were randomly selected from 12 census tracts in the BCQ MSA in which at least 7% of the population was born in Brazil, and in addition to having a non-response rate (56%) similar to government and other surveys of relatively hard-to-reach populations (Diffendal 2001), the weighted sample data suggest a population size of approximately 61,000 foreign-born Brazilian adults, and 3000 foreign- and U.S.-born born children. Furthermore, the estimated total number of foreign-born Brazilians residing in the BCQ MSA is statistically identical to that suggested by 2005–2009 American Community Survey data, as are their age and sex distributions. Information regarding migration and legal status, socioeconomic status, social capital, neighborhood characteristics and self-reported health behavior and conditions was collected from 307 foreign-born Brazilian adults and 120 of their U.S.- and foreign-born children. More detailed information about the BM-IHLSS study design and objectives have been published elsewhere (Holmes and Marcelli 2012, 2014; Marcelli 2014; Marcelli et al. 2009a). The authors have no conflicts of interest to report.

Measures

Serious psychological distress was measured using the K6 scale (Kessler et al. 2002), which includes six questions asking, “during the past 30 days, how often did you feel…” (1) sad; (2) nervous; (3) restless; (4) hopeless; (5) everything was an effort; and (6) worthless. The scale employs Likert-style responses ranging from (0) none of the time to (4) all of the time, for a possible cumulative value of 24. Scores of 13 or higher indicate serious psychological distress and were set equal to “1” while scores of 12 or below were set equal to “0.”

Neighborhood environment

The BM-IHLSS data employed two definitions of neighborhood—first, survey respondents were asked to evaluate the characteristics of “your neighborhood,” i.e., the boundaries were subjectively defined by the participants; second, we linked the BM-IHLSS data to block-level Summary File 1 (SF1) data from the 2000 census in order to estimate aggregate area-level characteristics. Length of residence was measured continuously as the number of months and years a participant had resided in his or her neighborhood. Neighborhood disorder was measured using a scale from 0 to 4 indicating whether respondents or their neighbors had experienced personal violence, had their homes broken into, had experienced property damage or had property stolen from them; a value of 0 means that the respondent had not experienced any of those things, while those with a value of 4 had experienced all. Social cohesion was also measured on a scale from 0 to 4, based on initial questions asking respondents to indicate to what extent they (0) strongly disagree, (1) disagree, (2) agree or (3) strongly agree (4) with four questions about the neighborhood environment, i.e., whether neighbors: (1) get along with each other; (2) are willing to help each other; (3) share the same values; and (4) know each other (Sampson and Raudenbush 1999). These variables were collapsed into dichotomous variables with a “1” set equal to “agree” or “strongly agree.” Therefore, a value of 0 indicated that respondents “disagree” or “strongly disagree” that their neighborhoods were cohesive on all measures and a value of four indicated strong perceived cohesion across all four measures. Population density was measured at the census block level as the number of residents per square mile. Percent homeownership was measured as the proportion of residents in each census block who owned their homes.

Demographic characteristics:

Age is a continuous variable indicating respondent years since birth, and sex is a dichotomous variable representing female (0) and male (1). Subject skin color was measured using the New Immigrant Survey Skin Color Scale (Massey and Martin 2003); respondents were shown a picture of ten human hands numbered 1–10 with the skin pigmentation growing increasingly darker from left to right along the scale, and asked to point to the hand that most resembled their own pigmentation.

Home and socioeconomic characteristics:

Marital status was set equal to “1” if respondents were married and “0” otherwise. Bachelor’s degree or higher is set equal to “1” if respondents had received at least a Bachelor’s degree and “0” otherwise. Earnings was a continuous measure tallying a subject’s earnings (in thousands of dollars) from all jobs in the previous year. Lastly, unauthorized legal status was set equal to “1” if the respondent was estimate to be unauthorized to reside in the United States and “0” otherwise.

Health status and behavior:

Body mass index was derived from biological measures of respondent height and weight, which were converted using the formula: BMI = kg/m². Obesity has been positively linked to psychological distress in several studies (Burdette and Hill 2008; Friedman et al. 2002; Hemmingsson 2014; Zhao et al. 2009), and we hypothesized serious psychological distress to be related to biological as well as social-emotional processes. Physical activity was set equal to “1” if the respondent engaged in at least 30 min of moderate exercise or 20 min of vigorous exercise in the previous week and “0” otherwise; smoking was set equal to “1” if the respondent reported currently smoking every day and “0” otherwise.

Statistical Analyses

Logistic regressions were performed using Stata 14. The svyset command was used to account for the complex nature of the survey data, including clustering and population weights. Five respondents were missing data for neighborhood disorder and 65 were missing BMI data. We therefore performed a multiple imputation procedure to adjust for the missing values and tested the models with and without the imputed results to ensure no significant bias was introduced. We first estimated psychological distress in the full sample (n = 307) using our neighborhood metrics and demographic characteristics. We then controlled in subsequent iterations for home and socioeconomic characteristics followed by health status and behavior.

Results

Descriptive Statistics

Table 1 below shows weighted sample characteristics for all adult Brazilians, and separated between those who did and did not exhibit SPD. Not shown are the summary statistics for the outcome, serious psychological distress, which had a confidence interval equal between 0.04 and 0.10 and a standard deviation of 0.26. Those with SPD had lived in their neighborhoods for a shorter amount of time (1.5 vs. 2.5 years), had greater neighborhood disorder (0.54 vs. 0.34 out of 4) and lower social cohesion scores (2.4 vs. 2.8 out of 4). They also had higher population density per square mile (26,536 vs. 22,969) and slightly higher homeownership rates (38% vs. 36%). Those respondents with SPD were also slightly older on average (36.4 vs. 33.4) and more likely to be male (56% vs. 54%). There were no notable differences in skin color between the two groups. Respondents with SPD were less likely to be married (42.0% vs. 56.4%) or to have been graduated from college (6% vs. 13%) and had lower earnings (~ $22,500 vs. $34,000). They were more likely to have been unauthorized to reside in the United States (82% vs. 71%). Additionally, those with SPD had a slightly higher BMI (27.1 vs. 25.3), substantially lower rates of physical activity (7% vs. 28%), and higher smoking rates (29% vs. 14%).

Table 1.

Weighted sample characteristics, 2007 BM-IHLSS (n = 307)

Variable	Total		Distress = 0		Distress = 1
	N = 307		N = 286		N = 21
	Weighted μ or %	SD	Weighted μ or %	SD	Weighted μ or %	SD
Neighborhood characteristics
Length of residence	2.42	2.25	2.49	2.29	1.51	1.29
Neighborhood disorder	0.36	0.80	0.34	0.79	0.54	0.89
Neighborhood social cohesion	2.78	1.10	2.81	1.09	2.37	1.06
Population density per sq mi	23,229	15,629	22,969	15,618	26,536	15,757
% Homeownership	0.36	0.21	0.36	0.21	0.38	0.23
Individual characteristics
Age	33.66	9.88	33.45	9.87	36.41	9.78
Male	0.54	–	0.54	–	0.56	–
Skin color	2.17	1.35	2.17	1.34	2.16	1.48
Home and socioeconomic characteristics
Married	0.55	0.50	0.56	–	0.42	–
Bachelor’s degree or higher	0.12	–	0.13	–	0.06	–
Annual earnings	33,000	22,965	34,055	23,499	22,491	9672
Unauthorized legal status	0.71	–	0.71	–	0.82	–
Health status and behavior
Body mass index	25.42	4.17	25.28	4.19	27.13	3.40
Physical activity	0.27	–	0.28	–	0.07	–
Current smoker	0.15	–	0.14	–	0.29	–

Logistic Regression Results

Table 2 shows the results of our logistic regression models estimating the relationships between neighborhood social cohesion and SPD for adult Brazilian migrants. In line with our main hypothesis, we found that neighborhood social cohesion was negatively and statistically associated with SPD in the final model. Specifically, for each unit increase on the social cohesion scale, there were 34% lower odds of having experienced SPD. While length of residence was inversely associated with having had SPD in the first two models, the effect was no longer significant after controlling for health status and behavior. Similarly, age, though positively associated with SPD in the first model, was no longer significant after adjusting for home and socioeconomic characteristics. The only other variable in the final model bearing a significant association to SPD was annual earnings; for every additional $2258 earned in the previous year, Brazilian migrants were 3% less likely to have had SPD.

Table 2.

Logistic regression of serious psychological distress on neighborhood characteristics, 2007 BM-IHLSS (n = 307)

	Model 1: neighborhood and demographic characteristics		Model 2: + socioeconomic characteristics		Model 3 + health status and behavior
	AOR	95% CI	AOR	95% CI	AOR	95% CI
Neighborhood characteristics
Length of residence	0.57	[0.37, 0.89]*	0.66	[0.44, 0.99]*	0.68	[0.46, 1.01]
Neighborhood disorder	1.36	[0.81, 2.30]	1.22	[0.67, 2.21]	1.27	[0.75, 2.14]
Neighborhood social cohesion	0.70	[0.49, 1.00]*	0.71	[0.50, 1.02]	0.66	[0.46, 0.94]*
Population density per sq mi	1.00	[1.00, 1.00]	1.00	[1.00, 1.00]	1.00	[1.00, 1.00]
Percent homeownership	4.96	[0.28, 88.51]	5.43	[0.29, 102.89]	3.02	[0.15, 61.21]
Individual characteristics
Age	1.05	[1.00, 1.11]*	1.06	[1.01, 1.11]	1.03	[0.98, 1.09]
Male	0.86	[0.31, 2.38]	0.95	[0.33, 2.73]	0.96	[0.23, 4.01]
Skin color	0.96	[0.64, 1.45]	0.90	[0.61, 1.33]	0.87	[0.56, 1.37]
Home and socioeconomic characteristics
Married			0.63	[0.18, 2.17]	0.70	[0.19, 2.58]
Bachelor’s degree or higher			0.50	[0.08, 3.08]	0.49	[0.06, 4.00]
Annual earnings			0.96	[0.94, 0.99]**	0.97	[0.93, 1.00]**
Unauthorized legal status			1.74	[0.60, 5.09]	1.67	[0.53, 5.27]
Health status and behavior
Body mass index					1.13	[0.96, 1.33]
Physical activity					0.21	[0.03, 1.37]
Current smoker					1.51	[0.33, 6.94]

^***p ≤ .001,

^**p ≤ .01,

^*p ≤ .05

Discussion

In this paper we utilized data from the 2007 Harvard-UMASS Boston Metropolitan Immigrant Health and Legal Status Survey (BM-IHLSS), a community-based migrant household probability sample of Brazilian migrants in the Boston–Cambridge–Quincy MSA, to investigate the relationships between neighborhood environment and serious psychological distress (SPD). Specifically, we hypothesized that neighborhood social cohesion is likely to protect against SPD. Our results demonstrate a strong negative association between neighborhood social cohesion and SPD among Brazilian adults. These relationships persist even after adjusting for sociodemographic characteristics and health status and behavior. This suggests that certain attributes of migrants’ neighborhoods, specifically those that foster a sense of community or place attachment, may indeed be protective of psychological health and especially important for studying in foreign born populations.

We also found that annual earnings were inversely associated with SPD, indicating that for every standard deviation increase in earnings (~ $2258), the odds of having SPD decreased by three percent. Although the effect size is small for this association, it is consistent across models. Furthermore, the majority of the Brazilian migrants in our sample were working in low income, precarious occupations, such as housecleaning and construction, and remitting large portions of their paychecks back to Brazil to support family members there (Marcelli et al. 2009a). Thus, even a small increase in annual earnings could have substantial impact on the psychological well-being of this group of migrants.

This study is subject to several limitations. First, the data are cross-sectional, so we cannot establish the causal direction of the observed associations. Although we conclude that neighborhood social cohesion protects against psychological distress, poor psychological health may also lead individuals to assess their neighbors in a less positive light. Second, our study population is unique, and although representative of Brazilian migrants in New England, it is not representative of the U.S. population as a whole. Finally, our neighborhood data are largely self-reported, with the exception of linked decennial Census data, and our analyses would likely benefit from additional objective data about the neighborhoods. However, these limitations leave the way open for future research on particular spatial, cultural and longitudinal associations between neighborhoods and psychological health among migrants and other populations.

Despite these study limitations, we are confident that our results underline an aspect of neighborhood environments that may be especially important for promoting mental health in vulnerable populations. In particular, our study highlights the important contribution of social cohesion in elaborating the manner in which neighborhoods may influence psychological health for migrants. Data collection programs that combine objective and subjective indicators of built and social environments along with objective measures of health have the potential to yield useful information for health policy formation and place-based interventions.