The Association between Neighborhood Social Cohesion and Depressive Symptoms in a Latino Sample and Differences By Language: The Multi Ethnic Study of Atherosclerosis and Neighborhood Study

Background: Neighborhood social cohesion (NSC) has been associated with depressive symptoms. However, studies on the association between NSC and depressive symptoms among Latinos and how the relationship may vary by sub-groups have been limited. English language proficiency may be an important characteristic as a proxy for acculturation and facilitator of social ties. Additionally, past research has found gender differences in the association between neighborhood features and depression. We assessed the association between NSC and depressive symptoms among Latinos and differences by language spoken at interview (English/Spanish), while accounting for neighborhood level differences. We hypothesized that NSC would be negatively associated with depressive symptoms and that the association would be amplified for those who spoke English at the interview compared to Spanish.

Methods: We used a Latino sample, ages 45 -84, from the Multi-Ethnic Study of Atherosclerosis and its ancillary Neighborhood Study (n=830); these cross-sectional data were collected 2010 - 2012. Participants were recruited from three sites (New York, NY; St. Paul, MN; and Los Angeles, CA) across 433 neighborhoods (census tract). First, the interclass correlation was calculated by adding depressive symptoms measured by the Center for Epidemiological Studies-Depression (CES-D, continuous) scale to a model with varying intercepts by neighborhood. Next, individual-level NSC was added to the model. In following models, individual level covariates (age, gender, income, nativity, Hispanic/Latino heritage) and a neighborhood level covariate (neighborhood socioeconomic status) were added. Finally, the individual level interaction term (NSC x language spoken at interview) was added to the fully adjusted models. Sensitivity analyses included rerunning models stratifying by gender and excluding participants using antidepressants.

Results: Participants were on average 69.2(SE=0.3) years old, 53% were women, and 60% had a household income of <$35,000. Fifty six percent of Mexican descent, 52% spoke English at interview, and reported a mean CES-D score of 9.5 (SE=0.3). In fully adjusted models, a one unit increase in NSC was associated with a 0.3 decrease in CES-D score (SE= 0.1, p= <0.001). Effect modification by language was not statistically significant. In fully adjusted models stratified by gender, the association between NSC and depressive symptoms was statistically significant only for women (b = -0.4, SE=0.1, p =<0.001). In the sample excluding participants with antidepressant use, the main effect in the fully adjusted model was no longer statistically significant when the interaction term (NSC x language spoken at interview) was added.

Conclusions: Consistent with the literature, NSC was negatively associated with depressive symptoms in this Latino sample. However, language did not modify the main effect. The association between NSC and depressive symptoms seems to be primarily driven by women, perhaps pointing to differences in how women interact with the neighborhood compared to men. This is consistent with past research that assessed gender differences in the effect of neighborhood characteristics on depression. Interpretation of these findings should take into consideration the cross-sectional dataset and therefore causality cannot be determined. Future research may consider exploring qualitatively the gender differences in exposure to neighborhood characteristics and its impact on mental health.