Introduction

Pregnancy is a particularly sensitive time for mother–offspring health. For the mother, there is a high risk for the onset of psychological distress, including depression, state anxiety, and pregnancy-related anxiety [1,2,3]. Poor maternal mental health can result in downstream consequences for fetal development and infant health, including preterm birth, low-birth weight, cognitive developmental deficits, and life-long non-communicable chronic disease risk [4,5,6,7,8,9,10,11,12,13]. Thus, investigating maternal mental health during pregnancy can have intergenerational public health implications.

The ecosocial model of health describes how distinct layers of social experiences can create population-level patterns of health disparities [14, 15]. Furthermore, people’s disparate experiences within a landscape of inequality can become embodied or “get under the skin” to expand minority health disparities [16, 17], which can have lasting intergenerational health effects [18]. Each of these interconnected levels of influence described by the ecosocial model can suggest strategies for intervention.

While community-level relationships, e.g., neighborhoods, have been studied extensively, much work has focused on poverty and deprivation [19,20,21,22,23,24,25]. Poverty and deprivation are significant points of influence, but studying only these issues limits our understanding of how a community functions. It is important to investigate the complex issues that prevent a community from thriving to create potential interventions toward improving the mental and physical health of individuals within said community. We make a unique contribution by expanding these questions of community-level impacts on mental health beyond only poverty or ethnic composition. Instead, we ask how a woman’s perceptions of the attitudes of her neighbors’ attitudes (henceforth “neighbor attitudes”) influence her mental health.

Subjective measures are critical here as there is growing evidence to suggest that perceptions over actual neighborhood characteristics may impose greater influence on mental health, particularly in minoritized families [22, 26,27,28,29,30]. In other work characterizing Latino mental health, scholars have shown that within Mexican American mother–offspring dyads, the subjective neighborhood, or how mothers perceive their neighborhoods and neighbors’ attitudes to them, relayed valuable information regarding their acceptance of them and their culture [29, 31]. Other aspects of neighborhoods, including cohesion and sociality, have also been shown to positively impact residents’ mental health [29, 32]. Together, these studies underscore the importance of considering subjective neighborhood and neighbor measures and their implications for health outcomes.

Latino Mental Health

Mental health disparities among racial and ethnic minority communities in the USA is a well-established issue. We focus on Latino mental health in the USA for several reasons. First, Latinos are one of the fastest-growing and largest ethnic minority groups in the USA (now second behind Asian Americans) [33]. Latinos tend to be disproportionately affected by various systemic, social, cultural, and economic barriers in the USA [34], which contributes to worse mental health outcomes [35]. Additionally, stressors are shown to contribute cumulatively to worsened mental health among reproductive-aged women, rather than a sensitized or habituated response [36]. The myriad of stressors affecting Latinos in the USA widens the minority mental health disparities in the USA. In parts of Los Angeles County, Latina mothers are most likely (25.9%) to be uninsured 6 months before becoming pregnant, as well as experience greater rates of fertility, preterm births, and low birthweights compared to mothers of other ethnic and racial groups [37]. Lastly, the Latino community faces many structural and cultural barriers that prevent them from receiving mental health treatment. Structural barriers include lack of health insurance or cultural miscommunication in health care, while cultural barriers include mental health stigma [38,39,40]. Only 31% of Latino adults with mental health disorders receive treatment each year compared to the USA, average of 45% [41]. Latina women experience rates of psychological distress 20% higher than non-Latina White women [42]. Latina women are also more likely to experience depressive symptoms, reportedly 27% higher compared to Latino men [39, 43]. These percentages vary by country of origin: US-born Latina women experience higher depression (44%) compared to foreign-born Latina women (34%) [44].

We focus on Latinos in Southern California because of its large Latino population compared to the rest of the USA (18.7% of the total population identifying as Latino in the USA compared to 40% of the total population in Southern California; [45]). Residents of Latin-American heritage represent 48% of the population in Los Angeles County and 34% of Orange County [45, 46]. In Los Angeles County, over one in three female residents are foreign-born, and of these residents, 56% are from Latin American countries [46].

Neighborhood Theoretical Frameworks

Latino mental health has been considered through the lens of social disorganization theory. Social disorganization theory [47, 48] hypothesizes that the quality of a neighborhood depends on three factors: poverty, residential stability, and ethnic composition. With regard to ethnic composition, social disorganization theory hypothesizes that higher concentrations of Latinos (i.e., ethnic homogeneity) promote resident mental health due to shared experiences [49]. Ethnic homogeneity lends itself to enabling trust among neighbors and increased social capital, thereby benefiting residents [50]. Consistent with the social disorganization theory, Latinos living in homogenous neighborhoods tend to have more social cohesion [32], better adolescent mental health [49, 51], and prenatal mental health [31] compared to ethnically heterogeneous neighborhoods. Notably, Curci and colleagues [31] found that a higher Latino neighborhood concentration promotes positive maternal mental health outcomes and child health outcomes among 322 Mexican American mother–offspring dyads from low-income neighborhoods in Phoenix, Arizona. While work from social disorganization theory has been influential in understanding Latino mental health, empirical evidence supporting this hypothesis is mixed [52]. White and colleagues [49], for instance, recently conducted a meta-analysis and found that children residing in less concentrated Latino neighborhoods actually fared better and had fewer mental health issues, contrary to the predictions of social disorganization theory. Thus, previous literature has inconsistencies regarding the relation between neighborhood ethnic composition and mental health among Latinos.

How these inconsistencies emerge stems partially from methodological and with it, conceptual issues pertaining to how homogenous neighborhoods are being measured. A large focus has been on constructs like foreign-born concentration, indexed by a composite percentage of Latino residents, or racial-ethnic concentration, indexed by the percent of Latino residents. These are imperfect assessments because these concentration indices could represent either homogeneous or heterogeneous neighborhoods as these estimates fail to account for factors like country of origin and geographic sampling bias [52]. The use of these constrained metrics likely stems from the gross unilateral categorizations that are often imposed on racial and ethnic minorities in the USA [53]. Latinos fall outside of the US system of racial categorization. Despite many health and population researchers acknowledging this limitation, there is no clear consensus or way forward for replacing this system of racial categorization.

Moreover, previous studies have conflicting results because neighborhood poverty and Latino concentration are highly correlated [52]. This is in part due to structural issues like redlining, gentrification, and/or racism that tend to systematically assort individuals into neighborhoods and thus create homogeneous neighborhoods. Refinements to social disorganization theory have been suggested by Kubrin and Weitzer [54] and White et al. [49, 52]. Notable suggestions include refraining from the use of concentration indices; increasing attention to the potential heterogeneity within a given group; and for the use of more culturally appropriate datasets that focus on the individual in order to capture more nuance.

Emerging work from segmented assimilation theory overlaps with social disorganization theory, in that higher concentrations of Latinos in a given neighborhood lend themselves to be “socially organized” due to a shared set of similar values, morals, and culture that creates relatively stable social cohesion. This tends to result in segmented groups that may not be a part of the hegemonic cultural group [49, 55]. Thus, there is less reliance on the socio-economic class and position of the residents and a higher emphasis on creating social structures and capital that allow its members to prosper. Importantly, segmented assimilation theory offers a more nuanced perspective in that they account for within-group cohesion, the idea that established groups might be able to offer benefits that are specific to their within-group members. It also recognizes that socially organized neighborhoods (i.e., established groups) can co-occur with other systemic challenges felt by the community, e.g., discrimination; [25, 52]. From this perspective, it acknowledges the fact that neighborhoods with high ethnic concentrations may be faced with higher forms of oppression, but that its residents may have developed behavioral strategies, e.g., disengagement, to mitigate the potential downsides of residing in these neighborhoods [52]. In addition, segmented assimilation theory recognizes how ethnic homogeneity can have negative downstream influences that can result in culture clashes between generations [55]. This is particularly salient when there are differences in expectations driven by societal and cultural norms (e.g., gender, family structure). Within this framework exists a stronger emphasis on relationships between neighbors, like shared values and social support, e.g., [56]. Focusing on these shared values can help encourage and support new forms of social capital that is needed for Latino communities to thrive.

Current Study

In the current study, we adopt a segmented assimilation perspective, which combined with the ecosocial model of health, can inform our understanding of reproductive and offspring health. The goal of this study is to address how a woman’s perceptions of her neighbors’ attitudes toward Latinos influence prenatal mental health during pregnancy. Through this framing, we aim to examine neighborhood quality in a way that goes beyond that of poverty and of ethnic composition. Specifically, we evaluate the relationship between perceptions of neighbor attitudes toward Latina culture and mental health among pregnant Latina women residing in Southern California. We focus on subjective measures of neighbor attitudes toward Latinos as this reflects how Latina mothers’ mental health might be influenced by the sociocultural environment of their neighborhood. We stratify our analyses by country of birth since previous work has shown different rates of affective disorders and mental health resources available to foreign-born and US-born women [57]. We predict that women who perceive their neighbors to be less accepting of them and their culture will experience greater levels of depression, state anxiety, and pregnancy-related anxiety as these feelings of rejection and low social support may exacerbate poor mental health. We are agnostic as to whether there will be any meaningful differences based on foreign-born status.

Methods

Cohort

This study is part of a larger research program, the Mothers’ Cultural Experiences (MCE) project, which aims to understand the health effects of maternal sociocultural experiences, and how those health effects are potentially transmitted to offspring during prenatal and early postnatal life. Eligible women were 18 years or older, fluent in either English or Spanish, and self-identified as Latina, Hispanic, Chicana, Mexican, and/or Latin American.

Data from an initial Wave 1 cohort, collected from 2016 to 2018 consisted of 361 pregnant and postpartum women recruited from medical clinics and breastfeeding classes in Los Angeles and Orange counties of Southern California. Thirteen women were deemed ineligible after the study ended. We omitted an additional 103 postpartum women because of our focus on the prenatal period and an additional nine women who did not indicate whether they were foreign-born or US-born, which was the criterion used to split the cohort. Thus, our final analytic cohort consisted of 239 women.

Material and Procedure

All materials and procedures were approved by the Institutional Review Boards of all participating institutions with appropriate reliances. This study adheres to the tenets of the Declaration of Helsinki. Women provided self-report data on their mental health, demographics, neighborhood perceptions, and their cultural orientation. We compensated them following the completion of the questionnaire, which was presented in either Spanish or English.

Variable Operationalization

Predictor Variable: Neighbor Attitudes

We utilized the Neighborhood Attitudes Toward Latinos scale [58] to determine the perceptions women have of their neighbors’ attitudes toward Latinos. This scale consisted of six items measured on a 5-point Likert scale (0 = Not at all true to 4 = Very true; e.g., “People in this neighborhood appreciate Latino culture and people”). Higher values on this scale indicate greater perceived neighbor acceptance toward Latino culture. We found that the reliability of this scale was low in our sample (Cronbach’s ɑ = 0.45). After conducting a confirmatory factor analysis, we found that the reliability model produced an adequate fit, χ2(5) = 10.79, p = 0.06; CFI = 0.97; TLI = 0.94, RMSEA = 0.08. However, in both the confirmatory factor analysis and when investigating the measure’s internal reliability, we found that the reverse-scored items did not significantly load onto the factor (Supplementary Materials). Thus, we removed the two reverse-coded items, and the resulting scale comprised four items, which exhibited better reliability (ɑ = 0.80). The Cronbach’s alpha for this modified scale was 0.83 for women taking the survey in English (henceforth ɑE) and 0.76 for women taking the survey in Spanish (henceforth ɑS).

Outcome Variables: Mental Health

We measure three mental health outcomes (depression, state anxiety, and pregnancy-related anxiety) with validated psychometric scales. For depression, we used the Edinburgh Postnatal Depression Scale (EPDS) [59, 60], which calculated depression scores based on 10 items. Each item was ranked on a 4-point Likert scale, and the sum was taken for a final score (see Supplementary Materials for more details). We found high reliability with this scale (overall ɑ = 0.84; ɑE = 0.86 and ɑS = 0.82). For anxiety, we used the Spielberger State-Trait Anxiety Inventory State Short-Form scale (STAI-SF) [61]. The STAI-SF consists of six items, three of which are reversed coded, and anchored on a 4-point scale (1 = Not at all to 4 = Very much). STAI-SF scores were averaged, and the complete range of possible scores was 1–4 (for an item breakdown, see Supplementary Materials). Again, we found good reliability (overall ɑ = 0.81; ɑE = 0.83 and ɑS = 0.77). For Pregnancy-Related Anxiety scale (PRA) [62, 63], each of the ten items was scored from 1to 4 and then the average taken (overall ɑ = 0.89; ɑE = 0.89 and ɑS = 0.90). We also intended to use the Perceived Stress Scale (PSS) as another mental health outcome variable, but due to low reliability (overall ɑ = 0.50) decided to forgo these analyses.

We treated the mental health variables as continuous in all models. This was due to Bowins’ [64] argument describing how mental health variables like depression tended to exist as a continuum in nature as opposed to discrete categories. We presented descriptive summary metrics of all mental health variables and the percentages of those with clinically significant symptoms of depression for each subset of the cohort (Table 1).

Table 1 Demographic and descriptive statistics of all mental health variables for each cohort
Full size table

Control Variables

In order to isolate the outcome variables of interest (i.e., depression, state anxiety, and pregnancy-related anxiety), we controlled for the other two mental health variables in the models (e.g., pregnancy-related anxiety models controlled for state anxiety and depression). We felt this was important since depression and anxiety comorbidities are common [65].

We also controlled for socio-economic status (SES) using a composite variable of subjective SES, education, and food insecurity developed by Fox [66]. This composite variable is made by standardizing each variable on a scale of 0–1 and then summing the three variables. This variable measures SES with the MacArthur Subjective Socioeconomic Status Scale [67], education with a single question of the highest level of education attained (i.e., “What is your highest level of education?”), and food insecurity with the two-item screen to identify families at risk for food insecurity [68]. We also controlled for relationship status and trimester in all models. These control variables each consistently affect prenatal mental health [8, 69], as well as neighbor attitudes [70].

Analytic Strategy

Missingness and Imputation

There was a total of 5% missingness across all variables. In order to increase the efficiency of our estimates, we ran multiple imputations by chained equations (MICE) [71] on our final analytic cohort (N = 239) and produced five imputed datasets. We tested the difference in missingness with relevant variables of interest, such as depression and foreign-born, and found nothing falsifying our missing at-random assumption (Supplementary Materials).

Models, Regression Diagnostics, and Data Analysis

We ran multiple linear regression models on all five imputed datasets. We performed Breusch-Pagan tests iteratively on each model among each of the five datasets. Depression and pregnancy-related anxiety models were consistently rejected by the Breusch-Pagan test, indicating heteroskedasticity in the data. As such, robust standard errors were calculated for all models. Variance inflation factor calculations also ensured low levels of multicollinearity in the models (Supplementary Materials).

We examined the effects of neighborhood attitudes of Latinos in three separate models, one for each mental health outcome, controlling for socio-economic status, relationship status, trimester, and other mental health variables. We ran the model separately for foreign-born and US-born women. All statistical analyses were performed in R statistical software, v4.0.3. We present pooled results below.

Results

Demographics

The first subset of the analytic cohort was composed of 108 foreign-born women (Mage = 30.90, SD = 1.21) who were primarily born in Mexico (78.6%), in relationships (89%), and experiencing high rates of food insecurity (40.5%). Using the clinical EPDS cutoff score for depression (score > 10), 16.8% of foreign-born women in our cohort had a high likelihood of at least minor clinical depression.

The second subset of the analytic cohort was composed of 131 US-born women (Mage = 27.70, SD = 5.51), also predominantly in relationships (87%) and experiencing high rates of food insecurity (38.0%). Using the clinical EPDS cutoff, 15.7% of US-born women in our cohort had a high likelihood of at least minor clinical depression. US-born and foreign-born women did not have statistically different mental health scores but had significantly different ages (Table 1).

Most women reported a large presence of Latinos in their neighborhoods (25.9% of women reported about half of their neighbors are Latino, and 45.2% of women reported more than half of their neighbors are Latino; Table 1). Only 21% of women in our cohort reported Latinos made up less than half of their neighborhood (Table 1). While there was some variation, these neighborhood trends are consistent with demographic patterns found in Los Angeles and Orange counties.

Results revealed that among foreign-born women, those who viewed their neighbors as having more positive attitudes toward Latino culture had significantly lower depression (pooled β =  − 0.70, SE = 0.29, p = 0.019) and lower PRA scores (pooled β =  − 0.11, SE = 0.05, p = 0.021), and greater state-anxiety scores (pooled β = 0.09, SE = 0.04, p = 0.021) (Table 2). There were no significant associations with any of the mental health outcomes among US-born women (Table 2).

Table 2 The relationship between social orientations and maternal mental health among foreign-born and US pregnant Latina
Full size table

Model comparison calculated from five imputed data sets against their respective null models produced the following pooled values among the foreign-born cohort (F-statistics; p values): EPDS (14.73; < 0.0001), STAI (12.87; < 0.0001), PRA (4.58; < 0.0001) and among the US-born cohort: EPDS (22.46; < 0.0001), STAI (19.91 < 0.0001), PRA (7.39; < 0.0001).

Discussion

We seek to integrate anthropological and sociological theories to explain mental health disparities among pregnant Latina women. The many interconnected levels of influence described by the ecosocial model suggest that health is not just our biology, but everything about the individual’s physical and socio-cultural environment. One of the exposures that can contribute to mother–offspring dyads is the role of neighborhoods. Thus, we find that a segmented assimilation perspective complemented the ecosocial model of health in that it provided a more nuanced application for measuring and accounting for neighborhood dynamics. Our focus on neighbor attitudes and their associations with maternal psychological distress is a first step to understanding intergenerational health among pregnant Latina women.

Our models found that foreign-born women who perceived their neighbors to have more positive attitudes toward Latinos tended to have lower levels of depression and pregnancy-related anxiety, but higher levels of state anxiety. That is, greater neighborhood acceptance of Latino people, language, or culture was associated with reduced psychological distress in two domains, and higher in one domain, only for our foreign-born cohort subset. These results are relatively consistent with other work demonstrating how neighborhood characteristics can negatively influence prenatal mental health [16, 31, 72, 73].

One possible explanation for the relationship between positive neighbor attitudes and better mental health may be attributed to the benefits of a positive social network. Among pregnant women living in Spain, social cohesion positively influenced mental health among only low and medium-SES groups [74]. This may be due to higher population density and the physical space, which lends itself to frequent interactions. Another possibility is that these women may reside in a neighborhood where their ethnic culture is preserved [75], resulting in a neighborhood that by default, will have a more positive view of Latinos. This harkens back to much of the discussed work on ethnic composition and social cohesion. However, van der Meer and Tolsma’s [76] review of this work does not find associations of ethnic diversity and interethnic social cohesion. Furthermore, there have been recent calls to re-evaluate how neighborhoods and ethnic concentrations are measured [53, 77]. Past studies tend to rely on the U.S. Census data, which often portrays an inaccurate assessment of racial/ethnic density. One of the main issues stems from the deeply rooted colonial narratives that treat ethnic and racialized minorities as discrete entities, White versus non-White. Additional confusion exists for Latinos who often feel unrepresented by the categories provided by the U.S. Census [53]. Thayer and colleagues [75] recommend the use of integrating both perceived and objective measures to assess neighborhood quality, which can be used in future research when examining neighborhood quality, social cohesion, and socio-cultural space.

It is important to note that state anxiety runs counter to our expected results and is inconsistent with previous work. It is possible that issues with the measure itself could have contributed to the mixed findings. Despite generally positive psychometric features of the STAI, scholars have criticized its use due to its inability to capture “pure anxiety” that is distinguishable from depression, thus, calling for better measures that accurately capture anxiety and its symptomatology [78]. Increased state anxiety among foreign-born women can signify challenges to conform to social norms within the neighborhood. Another plausible explanation for our findings that is unrelated to measurement concerns may stem from the pressure to acculturate and assimilate to the USA, e.g., [18, 79]. Statistical limitations related to power or confounding may also be responsible for unexpected findings (see Limitations below).

Our results revealed substantial differences between US-born and foreign-born women with regard to how perceptions of neighbors’ attitudes relate to prenatal mental health. Previous research highlights differences between US-born and foreign-born Latinos, including differences in social support networks, cultural values, mental health, and reasons for living in ethnic enclaves [39, 55, 57, 80]. These differences motivated us to separate the cohort by country of birth. However, we cannot say definitively why the results for US-born women were null. This may be due to the differences in cohort sample size. Alternatively, there may be distinct features about the US-born cohort, such as them being less sensitive to neighbor attitudes, but we will not speculate further.

We contribute to the literature by integrating a confluence of theoretical frameworks from anthropology, sociology, and ethnic studies to address the importance of neighbor attitudes on pregnant Latina mental health. We seek to move beyond strictly focusing on the ethnic composition of an individual’s neighbors. Additionally, it is critical to intergenerational health to ask these questions during pregnancy, thereby blending health disparity and sociology research frameworks. If neighbor attitudes do indeed play a role in pregnant Latina mental health, then this work could complement existing literature, and future work might benefit from investigating the heterogeneity of perceptions, experiences, and cultures within neighborhoods. The ultimate goal is to mitigate mental health disparities and inequities among Latinos.

Public Health Implications

Given the complexities of the incidence of depression and other mental health disorders within the Latino population, understanding the mechanism by which mental health disorders ensue in pregnant Latina women can help public health officials design interventions that target specific risk factors of mental health disorders. Prenatal maternal psychological distress (including stress, anxiety, and depression) has also been associated with low birth weight and pre-term birth [3, 9, 10, 81]. Low-birth weight and preterm birth are frequent causes of infant morbidity and mortality [82, 83], which in turn is implicated in the offspring’s long-term disease risk, including cardiovascular disease, obesity, diabetes, and psychopathology [82]. As a result, designing effective treatment and prevention plans for prenatal mental health is imperative within Latina communities.

Our findings suggest that public health interventions should promote diversity and cultural exchange at the neighborhood level in addition to improving aspects of neighborhood cohesion as part of maternal–fetal care management. We note here that increasing diversity does not necessarily mean that only members belonging to a racial or ethnic minority should be confined to a neighborhood. Rather, increasing diversity involves all members of a community and integrates various perspectives. To increase positive views of others and others’ cultures, one possible intervention could be to provide cultural education for the community [38]. Here, we underscore the importance of creating community cultural centers that can foster diversity, ethnic identity, and education programs. This type of intervention has the potential to reduce incidences of poor mental health among Latino adults and youth, cultivate coping skills for Latino adolescents who might face discrimination [44], and promote education for non-Latino community members who might be unfamiliar with Latino culture and values [38].

Limitations

We cannot comment on causality due to the observational design of our cross-sectional study. While we suggest, for example, that negative neighbor attitudes of Latinos may result in poorer mental health outcomes, the opposite causal direction may also be true: poor mental health may lead women to view their neighbors’ perceptions of Latinos through a negative lens.

This study is only among pregnant women. While women experience other pregnancy-related stressors apart from neighborhood strain, we can only statistically control for stressors that may confound the results. Modeling our predictions in cohorts of only US-born or foreign-born pregnant women helps control for the unique stressors of pregnancy and birthplace. We chose control variables that influence both neighbor attitudes and psychological distress. Our dataset is limited to the variables in the study. Future studies will benefit from larger datasets that include all possible variables.

Similarly, we did not have enough information about women’s mental healthcare access or whether they were already seeking treatment with a mental healthcare provider. It is possible that mental healthcare could influence the findings, possibly explaining some of the null results. Future research will also benefit from a comprehensive list of women’s mental health history, access to professional services, and treatment.

We ran a post-hoc sensitivity analysis for F-tests in G*Power (v3.1) to assess the strength of our regression models, setting our parameters to an alpha-error of 0.05. This revealed that our designed models can detect effect sizes (f 2) as small as 0.116 for foreign-born and 0.142 for US-born with 80% power. According to Cohen’s [84] guidelines where f 2 ≥ 0.02 is a small effect size and f 2 ≥ 0.15 is a medium effect size, we are powered to detect small effect sizes more so among the foreign-born cohort subset. The pattern of differences between our two cohorts may be attributable to these differences in sample sizes.

We also acknowledge potential issues with the term “Latina,” which inherently describes a very diverse population, including women of many different socio-cultural experiences and ethnicities. Our sample is mostly women who are of Mexican descent. Our findings, therefore, may have limited generalizability beyond pregnant Mexican American women living in Southern California.

Conclusion

Our findings contribute to the resurgent interest in work investigating the importance of perception of neighborhood attitudes and mental health among pregnant Latina women. We integrate theories and perspectives from anthropology, sociology, and health disparities research to examine how the wider social environment can influence mental health, particularly for foreign-born women, which can have implications for intergenerational embodiment. Future studies are needed to evaluate how the neighborhood context can improve maternal–fetal care management.