Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Nov 1.
Published in final edited form as: Aging Ment Health. 2023 May 26;27(11):2220–2228. doi: 10.1080/13607863.2023.2215180

Trajectories of Depressive Symptoms among Older African Americans: The Influence of Neighborhood Characteristics and Gender

Weidi Qin 1, Christy L Erving 2, Ann W Nguyen 3
PMCID: PMC10592451  NIHMSID: NIHMS1917993  PMID: 37231746

Abstract

Objectives:

Focusing on older African Americans, this study aims to 1) identify 9-year trajectories of depressive symptoms, 2) examine the association between baseline neighborhood characteristics (i.e. social cohesion and physical disadvantage) and trajectories of depressive symptoms, and 3) test whether the effects of neighborhood characteristics on depressive symptoms trajectories differ by gender.

Methods:

Data came from the National Health and Aging Trend Study. Older African Americans at baseline were selected (N = 1,662) and followed up for eight rounds. Depressive symptom trajectories were estimated using group-based trajectory modeling. Weighted multinomial logistic regressions were conducted.

Results:

Three trajectories of depressive symptoms were identified: “persistently low”, “moderate and increasing”, and “high and decreasing” (Objective 1). Objective 2 and 3 were partially supported. Specifically, high perceived neighborhood social cohesion was associated with a lower relative risk of being on the “moderate and increasing” versus the “persistently low” trajectory (RRR=0.64, p<0.01). The association between neighborhood physical disadvantage and depressive symptom trajectories was stronger among older African American men compared to women.

Conclusions:

High levels of neighborhood social cohesion may protect against increasing depressive symptoms in older African Americans. Compared to women, older African American men may be more vulnerable to negative mental health effects of neighborhood physical disadvantage.

Keywords: older Black Americans, depression, neighborhood context, gender difference, mental health

Introduction

Emerging evidence shows that structural-level factors, such as neighborhood characteristics in particular, have demonstrated effects on the levels of depressive symptoms (Andrews et al., 2021; Barr, 2018; Park, 2017; Ruiz et al., 2018). According to the Ecological Theory of Aging (Lawton, 1999; Scheidt & Norris-Baker, 2003), individuals’ appraisal of their environment context can influence their psychosocial responses and shape subsequent behaviors and health outcomes. This person-in-environment hypothesis provides a useful perspective in understanding how neighborhood characteristics may influence trajectories of depressive symptoms in older African Americans. Instead of focusing on individual-level factors alone, the person-in-environment association emphasizes that individual outcomes should be examined within a larger environmental context. Guided by the person-in-environment perspective, older African Americans are at higher risk of living in disadvantaged communities due to historical redlining, and thus may have more negative appraisals of their neighborhood and experience associated adverse health outcomes. For example, Black adults are four times more likely than other residents to live in neighborhoods with poverty rates at 40% or higher (Firebaugh & Acciai, 2016). However, little evidence is available to explain the influence of neighborhood characteristics on trajectories of depressive symptoms, which is relatively understudied in comparison to individual-level factors such as socioeconomic status (Kim, 2008). Therefore, it is important to understand how neighborhood factors are related to trajectories of elevated depressive symptoms in this group. Bridging this gap can elucidate modifiable structural-level factors influencing late-life depression among African Americans.

Late-life depression is a common mental health condition that can adversely impact older adults’ health and psychosocial well-being. Specifically, depression is associated with chronic physical health conditions, diminished social role functioning, increased healthcare utilization, and premature mortality (Kessler, 2012; Pratt & Brody, 2014; World Health Organization, 2021). It has been estimated that major depression affects approximately 1% to 5% of community-dwelling older adults (Centers for Disease Control and Prevention, 2021). Older adults with depression are more likely to have a worse trajectory and more chronic course of this condition than individuals in younger age groups (Haigh et al., 2018). According to a recent systematic review on the trajectories of depressive symptoms, the majority of studies have identified three to six distinct trajectory groups. These trajectory groups demonstrate two characteristics: severity (e.g., low, medium, and high) and stability (stable, increasing, and decreasing) (Musliner et al., 2016).

There has been a lack of evidence on trajectories of depressive symptoms specifically focusing on older African Americans. Prior population-based evidence shows that major depression affects approximately 7% of older African Americans (Woodward et al., 2013). A previous cross-sectional study has identified two latent profiles of depressive symptoms (i.e. low and high) among African American adults (Lincoln et al., 2007), and another study identified four longitudinal trajectories of depressive symptoms among African American adolescents (Repetto et al., 2004). However, within-group evidence on depressive symptom trajectories specifically among older African Americans is limited. Given that depression is a growing concern among older African Americans, a within-group examination of depressive symptom trajectories will provide a better understanding of the courses of depressive symptoms that are distinct to this group and inform culturally tailored mental health services.

The neighborhood can be a salient factor of depressive symptoms in older adults (Paczkowski & Galea, 2010), as older adults spend more time in their residential area than their younger counterparts, particularly when they have transitioned out of the workforce (Joint Center for Housing Studies of Harvard University, 2018). Some aspects of neighborhoods are protective of mental health for older adults. For example, a population-based study among older adults in England observed that a high level of neighborhood social cohesion (i.e., perceived mutual trust in the neighborhood) was associated with a slower increase in depressive symptoms compared to those reporting low social cohesion over 12 years (Ruiz et al., 2018). This study provides preliminary evidence that guides our study on neighborhood characteristics and trajectories of depressive symptoms. Additionally, among middle-aged and older African Americans in particular, Erving & Cobb (2021) found that contact with neighbors as well as neighborhood social group participation were associated with fewer depressive symptoms. In contrast, negative aspects of neighborhood, such as physical disadvantage, are associated with more depressive symptoms (Cho, 2022; Echeverría et al., 2008; Latkin & Curry, 2003). Older African Americans are more likely to reside in under-resourced and stress-inducing neighborhoods (Williams & Jackson, 2005). For example, recent evidence suggests that older African Americans perceive higher physical disadvantage and lower social cohesion in their neighborhoods compared to older white adults (Wong & Wang, 2022).

The association between neighborhood factors and depressive symptom trajectories may also differ by gender since older African American men and women experience unique intersectional vulnerabilities that give rise to depression in later life (Erving, 2022; Thomas Tobin et al., 2022). While evidence on the moderating effects of gender in the neighborhood-depression association is lacking specifically among older African Americans, prior research has reported differing findings among the overall older population. In particular, while some research has found that social cohesion has stronger effects on depressive symptoms among women but not men (Gutman & Sameroff, 2004), other findings suggest that poor neighborhood walkability (i.e. safe and convenient access to desired places in the neighborhood) is associated with more depressive symptoms among men but not women (Berke et al., 2007). Taken together, existing evidence indicates that social aspects of neighborhood may have more beneficial effects on older African American women’s mental health, while physical aspects of neighborhood may be more important to older African American men.

To address the research gap on depressive symptom trajectories among older African Americans and to clarify how neighborhood and gender interactively influence trajectory patterns, the objectives of the present study are to:

  1. identify 9-year trajectories of depressive symptoms among older African Americans;

  2. estimate the effects of neighborhood factors (i.e., social cohesion and physical disadvantage) on identified trajectories; and

  3. examine whether the associations between neighborhood factors and trajectories of depressive symptoms differ by gender.

Given the exploratory nature of these analyses, we are unable to provide specific hypotheses.

Materials and Methods

Sample

The study sample came from Round 1 (2011) to Round 9 (2019) of the National Health and Aging Trend Study (NHATS), a longitudinal panel study conducted annually among a nationally representative sample of US Medicare beneficiaries aged 65 or older (Montaquila et al., 2012). NHATS respondents were recruited in 2011 from the Medicare enrollment database by following a stratified three-stage sample design (Montaquila et al., 2012). The NHATS features an oversampling of African Americans and persons aged 85 and older, and contains rich data on biopsychosocial status, which is suitable for testing the study aims. At baseline, a total of 7,609 community-dwelling older adults responded to the survey. The inclusion criteria for the sample selection were participants who self-identified as African Americans (N = 1,662). The selected sample was followed up to capture 9-year trajectories of depressive symptoms. In the multivariable analyses, listwise deletion of respondents with missing data (N=158; 9.5%) on any of the predictor and control variables yielded a sample size of N = 1,504.

The baseline sample characteristics are presented in Table 3. Among the overall sample, about 80% of the study sample reported high social cohesion and about 29% reported the presence of physical disadvantage (i.e., negative physical features in the neighborhood). About 29% of study respondents aged between 65 to 69 (29%), 28% aged between 70 to 74, and slightly over 40% aged 75 or older. The majority were female (60%) and had high school education or below (65%). About one third lived alone (33%), and had possible or probable dementia (30%). About a quarter were hospitalized in the previous year (25%). On average, respondents reported 3 (range: 0 – 7) chronic conditions and a mean score of 1.19 for PHQ-2 (range: 0 – 6). Sample characteristics at each wave are presented in Supplementary Table 1.

Table 3.

Baseline Sample Characteristics by Trajectory Group of Depressive Symptoms

Overall
(N = 1,662)		 Persistently low
(N = 998)
 Moderate and increasing
(N = 553)		 High and decreasing
(N = 111)		 Range p value
High social cohesion (%, 95% CI) 79.6 (76.9, 82.0) 63.2 (59.4, 66.8) 30.6 (27.6, 33.8) 6.2 (4.7, 8.1) < 0.001
Presence of physical disadvantage (%, 95% CI) 28.8 (24.6, 33.5) 56.9 (52.0, 61.7) 36.8 (32.4, 41.3) 6.3 (4.0, 9.8) 0.158
Age (%) 0.159
 65–69 29.4 (27.6, 31.3) 63.3 (55.7, 7.02) 30.0 (23.4, 37.4) 6.8 (4.7, 9.7)
 70–74 28.4 (27.1, 29.7) 65.1 (60.4, 69.4) 30.6 (25.8, 36.8) 4.4 (3.0, 6.4)
 75–79 17.7 (15.7, 19.8) 54.7 (48.2, 61.0) 38.6 (32.6, 44.9) 6.7 (3.8, 11.4)
 80–84 13.6 (12.3, 15.1) 56.2 (50.6, 61.6) 35.1 (30.0, 40.5) 8.8 (5.9, 12.9)
 85–89 7.1 (6.1, 8.3) 56.6 (47.9, 64.9) 35.8 (26.8, 45.9) 7.6 (3.8, 14.6)
 90 and above 3.7 (3.2, 4.3) 59.1 (49.6, 68.0) 33.7 (25.1, 43.7) 7.2 (3.2, 15.2)
Gender (%, 95% CI) 0.281
 Male 39.8 (37.0, 42.6) 63.1 (58.7, 67.3) 31.2 (27.2, 35.23) 5.8 (4.3, 7.9)
 Female 60.2 (57.3, 63.0) 59.1 (54.4, 63.6) 34.1 (30.4, 38.0) 6.8 (5.1, 9.1)
Education (%, 95% CI) < 0.001
 Below high school 39.4 (36.0, 42.8) 52.2 (47.6, 56.7) 39.1 (35.3, 43.0) 8.8 (6.3, 12.0)
 High school graduate 25.2 (22.5, 28.1) 62.2 (56.7, 67.3) 32.1 (26.8, 38.0) 5.7 (3.8, 8.6)
 College, no degree 16.6 (14.3, 19.2) 63.8 (58.1, 69.2) 31.2 (24.8, 38.4) 5.0 (2.3, 10.3)
 College graduate or above 18.8 (15.7, 22.4) 74.2 (68.1, 79.4) 22.2 (16.7, 29.0) 3.6 (1.9, 6.6)
Income (%, 95% CI) < 0.001
 First quartile 24.3 (22.0, 26.8) 49.0 (42.4, 55.6) 43.2 (37.6, 49.1) 7.77 (4.9, 12.2)
 Second quartile 22.8 (20.6, 25.2) 55.4 (49.4, 61.1) 37.2 (31.4, 43.4) 7.4 (4.5, 12.2)
 Third quartile 25.4 (23.4, 27.5) 66.0 (61.0, 70.6) 26.4 (21.8, 31.6) 7.7 (5.3, 10.9)
 Fourth quartile 27.4 (24.7, 30.4) 70.5 (66.0, 74.7) 26.2 (22.1, 30.8) 3.2 (2.1, 4.9)
Living alone (%, 95% CI) 33.4 (30.5, 36.5) 59.4 (54.4, 64.3) 33.8 (28.9, 39.0) 6.8 (4.9, 9.4) 0.791
Dementia status (%, 95% CI) < 0.001
 No dementia 70.4 (68.5, 72.3) 65.5 (61.4, 69.4) 29.1 (25.5, 33.1) 5.4 (4.1, 7.0)
 Possible dementia 14.2 (12.3, 16.4) 52.5 (44.4, 60.4) 38.6 (31.7, 45.9) 9.0 (6.4, 12.5)
 Probable dementia 15.3 (13.9, 16.9) 46.0 (40.9, 51.2) 45.0 (38.9, 51.3) 9.0 (5.8, 13.6)
Self-care activity disability (mean, 95% CI) 6.23 (6.07, 6.39) 5.74 (5.56, 5.93) 6.69 (6.35, 7.04) 8.50 (7.75, 9.24) 4 – 16 < 0.001
Household activity disability (mean, 95% CI) 8.65 (8.40, 8.91) 7.70 (7.42, 7.98) 9.67 (9.23, 10.11) 12.46 (11.28, 13.64) 5 – 20 < 0.001
Past-year hospitalization (%, 95% CI) 25.3 (22.9, 27.8) 47.6 (43.5, 51.8) 42.5 (38.4, 46.7) 9.9 (7.5, 12.9) < 0.001
Count of chronic conditions (mean, 95% CI) 2.54 (2.47, 2.62) 2.36 (2.29, 2.44) 2.75 (2.59, 2.91) 3.22 (2.84, 3.60) 0 – 7 < 0.001
PHQ-2 score (mean, 95% CI) 1.19 (1.12, 1.27) 0.54 (0.48, 0.59) 2.00 (1.86, 2.14) 3.23 (2.83, 3.63) 0 – 6 < 0.001
Proxy respondents (%) 10.2 48.5 37.3 14.2 < 0.001
Attrition rate by 2019 (%) 69.6 60.4 32.3 7.3 0.219
Trajectory group Proportion (%, 95% CI) - 60.7 (57.1, 64.1) 32.9 (29.8, 36.2) 6.4 (5.1, 8.0) -
Open in a new tab

Note. PHQ-2 = Patient Health Questionnaire. CI = confidence interval. NHATS-provided analytic sampling weights and survey design factors were applied to generate population estimates. All percents and means were weighted except for proxy and attrition status. P values indicated the significance level of chi-square tests for categorical variables and t-tests for continuous variables. Range was reported for continuous variables.

Measures

Depressive symptoms.

Depressive symptoms were assessed by a two-item Patient Health Questionnaire (PHQ-2), a previously validated and commonly utilized brief depression screener among older adults (Li et al., 2007). The PHQ-2 asked the respondents to rate on two items: over the last month, how often have you 1) had little interest or pleasure in doing things, and 2) felt down, depressed, or hopeless? The response categories for each item ranged from 0 (not at all) to 3 (nearly every day). The two items were summed up with a total score ranging from 0 to 6. Previous evidence suggested that a cut-off score of 3 or more on the PHQ-2 indicated elevated depression (Li et al., 2007). Therefore, we dichotomized the PHQ-2 score into 0 (not depressed) and 1 (elevated depressive symptoms). As PHQ-2 does not capture the multidimensionality of depressive symptoms, it was not used to diagnose major depression. Instead, we used the scale to indicate the level of depressive symptoms.

Neighborhood characteristics.

Neighborhood social cohesion was measured using a three-item scale. Specifically, respondents were asked to rate to what extent they agree with the following statements: 1) people in the community know each other well, 2) people are willing to help each other, and 3) people can be trusted. Each item was rated on a three-point scale, which were 1 = do not agree, 2 = agree a little, 3 = agree a lot. Scores were averaged across all three items to create an indicator of social cohesion (Cronbach’s α=0.746). Based on previous utilization of this scale (Latham & Clarke, 2016; Qin et al., 2022), a dichotomized indicator of “low cohesion” (0 = low cohesion, 1 = high cohesion) was computed for the lowest 15th percentile of the distribution (score of 1.67 or below).

Neighborhood physical disadvantage was assessed with a four-item environmental checklist, which was completed by the interviewers based on their observations on the neighborhood environment. The items described the presence of: 1) litter, broken glass, or trash on sidewalks and streets, 2) graffiti on buildings and walls, 3) vacant or deserted houses or storefronts, and 4) houses with foreclosure signs. Responses were recorded on a four-point scale (1 = none, 2 = a little, 3 = some, 4 = a lot). Scores were averaged to create an indicator of neighborhood physical disadvantage (Cronbach’s α=0.769). Because the presence of any physical disadvantage was observed for about 30% of the study sample, and the distribution was highly skewed, physical disadvantage was dichotomized to capture any sign of disadvantage (score of 2 or above; 0 = no disadvantage, 1 = presence of disadvantage). This utilization was consistent with previous research (Latham & Clarke, 2016; Qin et al., 2022).

Both neighborhood measures were derived from a previous validated measurement study with oversampling of older African Americans (Cagney et al., 2009). By using self-reported social cohesion and interviewer-reported physical disadvantage in the neighborhood, we were able to examine both subjective and objective aspects of neighborhood quality.

Covariates.

Covariates were selected based on previous research on depressive symptom trajectories (Montagnier et al., 2014; Xiang, 2019). These variables were controlled in multivariable analyses to limit their potential confounding effects on depressive symptoms and estimate the independent effects of neighborhood characteristics in testing Objective 2 and 3. We controlled for demographics including age (65–69, 70–74, 75–79, 80–94, 85–89, 90+ years) and gender (male or female). Socioeconomic status variables included education (below high school, high school graduate, some college but no degree, college graduate or above) and family income (in US dollars in 2011). Family income was coded in quartiles for the regression analyses. We also adjusted for respondents’ living arrangement (living alone vs. living with someone). Additionally, we included an indicator of dementia using NHATS-provided classifications (no dementia, possible dementia, probable dementia). Dementia status was classified into three categories (no dementia, possible dementia, or probable dementia) based on cognitive tests, the Ascertain Dementia 8 Questionnaire Screening Interview (AD8), and self-report of a physician diagnosis (Kasper et al., 2013). At baseline, 129 participants had a diagnosis of dementia. Indicators of physical functioning were limitations in self-care activities (i.e. activities of daily living) and household activities (i.e. instrumental activities of daily living). Self-care activity was assessed in four domains: eating, dressing, toileting, and bathing. Household activity was assessed in five domains: laundry, grocery shopping, preparing hot meals, financial management, and taking medications. Previous research has created a summary score for limitations in self-care and household activity, respectively, to indicate the level of disability (Freedman et al., 2013; Gill & Williams, 2017). To further account for health, we controlled for past-year hospitalization and a count of self-reported physician-diagnosed chronic conditions (hypertension, heart disease or heart attack, arthritis, osteoporosis, diabetes, lung disease, stroke, and cancer). We also controlled for the baseline PHQ-2 score because prior evidence shows that baseline depression may be associated with depressive symptom trajectories among older adults (Montagnier et al., 2014; Xiang, 2019). The socio-demographic and health variables were not highly collinear with neighborhood variables (Supplementary Table 2).

Analysis Strategy

The analyses involved two parts. In part one, trajectories of elevated depressive symptoms during nine waves of the study were identified using group-based trajectory modeling extended to account for non-random attrition (Haviland et al., 2011), which is a type of finite mixture modeling utilized to identify groups of individuals following similar progressions of depressive symptoms over time. A logit function was used to model attrition simultaneously with the trajectory group membership as a function of time before the loss to follow up. Probabilities of dropout and trajectory group membership were assumed to be independent. The traj command in Stata were used to perform group-based trajectory modeling (Jones & Nagin, 2013). The model selection was based on the following criteria: 1) changes in Bayesian Information Criteria (BIC) and Log Bayes Factor; 2) close correspondence between each trajectory group’s estimated membership probability and the observed proportion of the study sample classified to that group based on the largest posterior probability of membership; 3) the odds of correct classification (i.e., average posterior probability > 0.7; 4) no less than 5% of the sample were assigned to a trajectory group; and 5) conceptual consideration on whether each group was distinct (Nagin, 2010). Respondents were assigned to the trajectory group to which they had the largest posterior probability of membership. After identifying the trajectory groups of depressive symptoms, we conducted descriptive analyses to capture the sample characteristics by identified trajectory groups.

In part two, three weighted multinomial logistic regressions were conducted to examine neighborhood characteristics and trajectory groups identified in part one (Montagnier et al., 2014). Consistent with procedures in previous research (Killeen et al., 2022; Xiang, 2019; Xiang & Cheng, 2019), multinomial logistic regression was utilized to predict trajectory group of depression. In Model 1, baseline neighborhood characteristics and covariates were entered to estimate the effects of baseline neighborhood social cohesion and physical disadvantage on the trajectories of depressive symptoms. In Model 2 and Model 3, two interactions, “social cohesion × gender” and “physical disadvantage × gender” were entered respectively to test the moderating effects of gender in the relationship between neighborhood factors and the trajectories of depressive symptoms. Taylor linearization was used for variance estimation to generate weighted results and match population estimators.

Data from proxy respondents were included and adjusted for two reasons. First, including proxy responses can reduce selection bias and maintain sample size (Skolarus et al., 2010), since proxy respondents represent an older and less healthy group. Second, statistically controlling for proxy status can decrease the bias induced by proxy data (Wolinsky et al., 2012). We also performed sensitivity analyses among non-proxy respondents to test the robustness of the study findings. All analyses were adjusted for NHATS-provided analytic weights to generate weighted estimates. All analyses were performed in Stata 17 (StataCorp, 2021).

Results

Nine-Year Trajectories of Depressive Symptoms

For study objective 1, a logit model with three trajectories of depressive symptoms was found to best fit the nine-year rounds of data based on Bayesian Information Criterion (BIC, Table 1), group proportion larger than 5%, group distinctiveness, and the average posterior probability of group assignments larger than 0.70 (Table 2). The three identified trajectory groups were shown in Figure 1. Group 1, labelled as “persistently low”, indicated a low risk of having elevated depressive symptoms over the 9-year study period, consisting of 60.7% of the weighted sample. Group 2 was labeled as “moderate and increasing” as the probability of having elevated depressive symptoms was moderate and showed a slightly increasing trend. This group represented about 32.9% of the weighted sample. Group 3 captured respondents who had “high and decreasing” probability of elevated depressive symptoms (6.4% of the weighted sample).

Table 1.

Model Selection Using Bayesian Information Criterion (BIC), Log Bayes Factor, and Estimated Group Proportions in Determining the Number of Trajectory Groups for Elevated Depressive Symptoms

Number of Groups BIC Log Bayes Factor Group 1 (%) Group 2 (%) Group 3 (%) Group 4 (%) Group 5 (%)
1 −3768.54 - 100.00 - - - -
2 −3550.59 217.95 66.38 33.62 - - -
3 −3550.82 −0.23 48.32 40.79 10.89 - -
4 −3569.80 −18.98 20.98 31.77 37.32 9.93 -
5 −3589.76 −19.96 26.40 1.72 17.94 44.42 9.52
Open in a new tab

Note. Log Bayes Factors were estimated using 2(∆BIC). The model solution with 4 groups failed to meet the criteria for acceptable average posterior probability.

Table 2.

Model Diagnosis for Selected Group-Based Trajectory Model

Group Estimated Group Membership Observed Group Membership Average Posterior Probability
1 47.6% 60.7% 70.8%
2 42.2% 32.9% 71.9%
3 10.2% 6.4% 71.7%
Open in a new tab

Note. The acceptable average posterior probability for each group was larger than 70%.

Figure 1.

Figure 1.

Open in a new tab

Trajectories of elevated depressive symptoms over 9 years, showing the predicted probability of having elevated depressive symptoms at each survey round. Weighted proportions among the population for each trajectory group were shown in parentheses.

The trajectories of depressive symptoms differed significantly across individual-level socio-demographics and health conditions (Table 3). Specifically, respondents in the “persistently low” group had a greater probability of living in neighborhoods with high social cohesion, had a college degree or above, had higher incomes, had no dementia, no self-care and household activity disability, no past-year hospitalization, fewer chronic conditions, and lower score of the PHQ-2.

Neighborhood Characteristics and Depressive Symptom Trajectories

For study objective 2, results from weighted multinomial logistic regression models are presented in Table 4. The reference group was the “persistently low” trajectory of depressive symptoms. Model 1 shows the main effects of neighborhood characteristics in predicting trajectories of depressive symptoms, controlling for the effects of socio-demographic and health covariates. Results show that high baseline social cohesion in the neighborhood was associated with lower relative risk of being on the “moderate and increasing” versus the “persistently low” trajectory group (relative risk ratio [RRR] = 0.64, 95% Confidence Interval [CI]: 0.46 – 0.89). However, no significant associations were observed between neighborhood physical disadvantage and the relative risk of being in the “high and decreasing” versus “persistently low” group. Additionally, the highest income quartile was associated with a lower risk of being in the “moderate and increasing” trajectory group, whereas those being hospitalized in the past year were more likely to be in this group. Having probable dementia was linked to a lower risk of being in the “high and decreasing” group. Also, a higher PHQ-2 score at baseline was associated with increased risks of being in both “moderate and increasing” and “high and decreasing” trajectory groups. Sensitivity analyses among non-proxy respondents conformed to the main analyses.

Table 4.

Results of Multinomial Logistic Regressions Predicting Trajectories of Elevated Depressive Symptoms (N = 1,504)

Model 1 Model 2
Baseline predictors (a) Moderate and increasing versus persistently low (b) High and decreasing versus persistently low (a) Moderate and increasing versus persistently low (b) High and decreasing versus persistently low
RRR (95% CI) p value RRR (95% CI) p value RRR (95% CI) p value RRR (95% CI) p value
High social cohesion 0.64 (0.46, 0.89) 0.008 0.91 (1.49, 1.67) 0.760 0.63 (0.45, 0.88) 0.006 0.89 (0.48, 1.64) 0.717
Presence of physical disadvantage 0.95 (0.69, 1.30) 0.732 0.64 (0.36, 1.16) 0.144 1.56 (0.94, 2.57) 0.083 1.34 (0.52, 3.50) 0.547
Presence of physical disadvantage × female - - - - 0.43 (0.23, 0.82) 0.010 0.29 (0.09, 0.95) 0.041
Age (ref=65–69)
 70–74 1.05 (0.70, 1.58) 0.824 0.84 (0.38, 1.86) 0.677 1.07 (0.71, 1.62) 0.742 0.87 (0.40, 1.92) 0.732
 75–79 1.46 (0.93, 2.67) 0.097 1.10 (0.45, 2.67) 0.839 1.47 (0.94, 2.30) 0.090 1.11 (0.46, 2.70) 0.816
 80–84 1.12 (0.70, 1.82) 0.631 1.24 (0.55, 2.78) 0.600 1.18 (0.73, 1.91) 0.487 1.34 (0.60, 2.99) 0.474
 85–89 0.94 (0.51, 1.71) 0.835 0.65 (0.22, 1.93) 0.442 0.96 (0.52, 1.76) 0.894 0.68 (0.23, 2.01) 0.482
 90 and above 0.98 (0.49, 1.98) 0.960 0.74 (0.20, 2.80) 0.661 1.01 (0.50, 2.04) 0.976 0.79 (0.21, 2.94) 0.724
Female (ref=male) 0.92 (0.66, 1.27) 0.601 0.72 (0.41, 1.25) 0.237 1.17 (0.80, 1.73) 0.413 1.01 (0.54, 1.92) 0.967
Education (ref=below high school)
 High school graduate 0.80 (0.55, 1.16) 0.240 0.58 (0.29, 1.14) 0.116 0.81 (0.55, 1.19) 0.286 0.58 (0.29, 1.15) 0.120
 College, no degree 0.73 (0.46, 1.16) 0.187 0.40 (0.16, 1.01) 0.052 0.75 (0.47, 1.19) 0.220 0.40 (0.16, 1.03) 0.057
 College graduate or above 0.67 (0.43, 1.05) 0.081 0.45 (0.20, 1.07) 0.072 0.69 (0.44, 1.09) 0.116 0.46 (0.19, 1.12) 0.087
Income (ref=first quartile)
 Second quartile 1.00 (0.67, 1.49) 0.991 1.42 (0.68, 2.95) 0.347 1.00 (0.67, 1.49) 0.991 1.41 (0.68, 2.93) 0.353
 Third quartile 0.68 (0.44, 1.05) 0.082 1.51 (0.69, 3.30) 0.301 0.67 (0.43, 1.04) 0.074 1.53 (0.71, 3.30) 0.275
 Fourth quartile 0.58 (0.36, 0.94) 0.026 0.64 (0.26, 1.61) 0.345 0.58 (0.36, 0.93) 0.023 0.64 (0.25, 1.59) 0.335
Living alone 1.06 (0.78, 1.44) 0.698 1.41 (0.80, 2.48) 0.230 1.04 (0.76, 1.41) 0.807 1.37 (0.78, 2.40) 0.273
Dementia status (ref: no dementia)
 Possible dementia 1.07 (0.70, 1.64) 0.751 1.03 (0.44, 2.37) 0.951 1.05 (0.68, 1.62) 0.810 1.01 (0.44, 2.33) 0.973
 Probable dementia 1.14 (0.69, 1.86) 0.613 0.31 (0.11, 0.90) 0.031 1.09 (0.66, 1.79) 0.737 0.30 (0.11, 0.86) 0.025
Self-care activity disability 1.00 (0.93, 1.07) 0.961 1.08 (0.96, 1.21) 0.191 1.00 (0.93, 1.07) 0.943 1.07 (0.96, 1.20) 0.230
Household activity disability 1.01 (0.97, 1.06) 0.582 1.07 (0.99, 1.16) 0.095 1.01 (0.97, 1.06) 0.616 1.07 (0.96, 1.20) 0.097
Past-year hospitalization 1.50 (1.08, 2.07) 0.015 1.43 (0.81, 2.54) 0.218 1.53 (1.10, 2.12) 0.011 1.47 (0.83, 2.60) 0.187
Count of chronic conditions 1.04 (0.93, 1.16) 0.462 1.18 (0.96, 1.45) 0.120 1.04 (0.94, 1.17) 0.432 1.18 (0.96, 1.45) 0.111
PHQ-2 score 2.44 (2.15, 2.77) < 0.001 3.48 (2.87, 4.23) < 0.001 2.48 (2.19, 2.82) < 0.001 3.56 (2.93, 4.34) < 0.001
Proxy respondents 0.77 (0.42, 1.41) 0.394 1.89 (0.60, 6.02) 0.279 0.78 (0.42, 1.43) 0.416 1.89 (0.62, 5.79) 0.265
Open in a new tab

Note. RRR = relative risk ratio. CI = confidence interval. PHQ-2 = Patient Health Questionnaire. Reference trajectory group was “persistently low”. NHATS analytic weights were applied to generate population estimates. Findings on non-significant interaction term of “neighborhood social cohesion × gender” were presented in Supplemental Table 3.

Neighborhood Characteristics and Depressive Symptom Trajectories by Gender

For study objective 3, an interaction term between “gender × presence of physical disadvantage” in Model 2 was significant in predicting the relative risk of being in the “moderate and increasing” trajectory (RRR = 0.43, 95% CI: 0.23 – 0.82). This interaction remained significant in sensitivity analysis among non-proxy respondents (RRR = 0.43, 95% CI: 0.23 – 0.83). To facilitate interpretation, the interaction was plotted in Figure 2. Specifically, in neighborhoods with the presence of physical disadvantage, older African American men had higher relative risk than women of being on the “moderate and increasing” versus the “persistently low” trajectory group, whereas no gender difference was found in neighborhoods with no presence of physical disadvantage. Although a marginally significant interaction between “gender × presence of physical disadvantage” was found for “high and decreasing” versus “persistently low” group (RRR = 0.29, 95% CI: 0.09 – 0.95), this interaction became non-significant in the sensitivity analyses among non-proxy respondents (RRR = 0.29,95% CI: 0.07 – 1.11). No significant interaction was observed between gender and high social cohesion (Supplementary Table 3).

Figure 2.

Figure 2.

Open in a new tab

Significant difference in effects of gender on predicted probability of depressive symptom trajectory of “moderate and increasing” versus “persistently low” by the status of neighborhood physical disadvantage among older African Americans, controlling for baseline socio-demographics and health status. Figure based on Table 4 Model 2(a).

Discussion

The present study is among the first to identify the long-term trajectories of elevated depressive symptoms among nationally representative older African Americans, and to understand the interactive effects of neighborhood factors and gender on these trajectories. Three primary findings emerged. First, there are likely three distinct depressive symptom trajectories among older African Americans: “persistently low”, “moderate and increasing”, and “high and decreasing”. Second, high perceived neighborhood social cohesion may protect older African Americans from “moderate and increasing” relative to “persistently low” depression trajectories. Third, the positive association between neighborhood disadvantage and risk of “moderate and increasing” depressive symptoms is stronger among men than women.

Our study adds to the literature by identifying trajectories of depressive symptoms that are specific to older African Americans. The number of distinct trajectory groups identified in the study sample (i.e. 3 groups) is consistent with previous findings based on samples with predominantly older White adults. In particular, prior studies have identified three to six groups, with the majority of individuals in the group characterized by persistently low or no depressive symptoms (Byers et al., 2012; Kuchibhatla et al., 2012; Montagnier et al., 2014; Musliner et al., 2016). Similarly, the largest proportion of the present study sample of older African Americans (60.7%) followed a trajectory of persistently low depressive symptoms. Nevertheless, our findings also differ from previous research that identified four or more trajectory groups (Killeen et al., 2022; Musliner et al., 2016). Specifically, we did not identify a trajectory group characterized by persistently high depressive symptoms, which was found among samples with primarily older White adults. One possible explanation is that while older African Americans are exposed to many stressful life events, they also have high levels of coping resources, resulting in lower prevalence of clinically significant depression relative to their White counterparts (Louie et al., 2021). Given that the existing literature on within-group examinations of depressive symptom trajectories among African Americans has focused on adolescents (Campbell-Grossman et al., 2016; Repetto et al., 2004), the present study extended the literature by adding to our understanding of long-term trajectories among older African Americans.

Expectedly, neighborhood characteristics shaped depressive symptoms trajectories. Our findings show that neighborhood social cohesion was associated with lower risk of being in the “moderate and increasing” compared to the “persistently low” trajectory group, suggesting that older African Americans who perceive strong community support and a trusting network among neighbors are less likely to follow a trajectory characterized by increasing depressive symptoms over time. Theoretically, the finding adds evidence for the Ecological Theory of Aging, which argues that environmental and structural factors can shape older adults’ health and mental health trajectories (Scheidt & Norris-Baker, 2003). In particular, positive appraisals of one’s neighborhood environment may indicate strong community-level social support that has protective effects against depressive symptoms among older African Americans. Alternatively, neighborhood social cohesion may serve as a psychosocial coping factor that precludes negative mental health outcomes. Empirically, this finding is consistent with prior evidence among older Korean women (Park, 2017). Likewise, a previous study among older adults in England reported that although individuals perceiving high social cohesion had increased depressive symptoms, symptoms increased at a lower rate compared to those reporting low social cohesion (Ruiz et al., 2018). These findings highlight that neighborhood social cohesion may be a particularly important protective factor that can influence trajectories of depressive symptoms among older African Americans. However, we also found that neighborhood social cohesion did not influence the risk of being in the depressive symptom trajectories group characterized by a high and decreasing pattern. Earlier research found that individuals following a persistently high depressive symptoms were more likely to be diagnosed with clinically significant depression and major depressive disorder (Musliner et al., 2016). Therefore, trusting relationships with neighbors may be less salient in protecting older African Americans who chronically suffer from severe depression.

Although no main effects of neighborhood physical disadvantage were identified, gender was a robust moderator in the association between physical disadvantage and the risk of “moderate and increasing” depressive symptoms. Specifically, among individuals living in neighborhoods with the presence of physical disadvantage, older African American men were more likely to follow a moderate and increasing trajectory of depressive symptoms than older African American women. Similar to previous evidence that better physical environments within neighborhoods was associated with reduced depressive symptoms in older men but not older women (Berke et al., 2007), our findings suggest that older African American men may be more vulnerable to adverse mental health consequences of neighborhood physical disadvantages. This gender difference may be explained by gender-specific stress effects. Specifically, while African American men and women experience common social stressors that negatively affect their mental health, some stressors are distinct to African American men. In related research, recent evidence suggested that perceived neighborhood crime was associated with depressive symptoms only among older African American men but not among women, possibly due to society-level criminalization of Black men (Erving, 2022). As the negative physical conditions of neighborhoods (i.e. graffiti, littering, deserted houses, and foreclosures) can signal crime-related safety concerns (Barnett et al., 2018), these neighborhood conditions may be more pronounced risks for depression among African American men than women. Nevertheless, we did not find moderating effects of gender on social cohesion and trajectories of elevated depressive symptoms. The non-significant finding may be explained by previous findings that there is no gender difference in social cohesion (Choi et al., 2015). It is possible that social cohesion, as a subjective perception of one’s neighbors, influence mental health trajectories among both black men and women. Additionally, our findings also point to individual-level factors influencing the trajectories of depressive symptoms.

Additionally, our study reveals significant covariate effects. Older African Americans with poorer health status have a greater probability of following a trajectory characterized by increasing depressive symptoms over time. Specifically, individuals who have probable dementia or are hospitalized in the past year may be at heightened risk of subsequent depressive symptoms, suggesting that physical and cognitive health may also play a role in trajectories of depressive symptoms. Echoing the Ecological Theory of Aging (Lawton, 1999; Scheidt & Norris-Baker, 2003), our findings highlight both neighborhood-level and individual-level risk factors that may contribute to late-life depressive symptoms.

Limitations and Strengths

Study findings should be interpreted in light of the limitations. First, despite being validated in previous research, the PHQ-2 is not a diagnostic measure of depression and is subject to recall bias and ceiling effects. The PHQ-2 also did not capture the multidimensionality of depressive symptoms. Future research could consider using measures that capture a wider spectrum of depressive symptoms, such as PHQ-9 or the Center for Epidemiologic Studies Depression Scale (CES-D). Second, a small proportion of the study sample was classified into the trajectory group of high and decreasing depressive symptoms, which may lead to underpowered regression analysis to identify the association between neighborhood variables and depressive symptom trajectories. Third, the spatial definition of neighborhood is ambiguous for the scales of social cohesion and physical disadvantage, as respondents’ perception of community size may differ. Additionally, some important features of physical disadvantage, such as barrier-free sidewalks and street connectivity, were not captured, which may lead to underestimation of the association between the physical environment and depressive symptom trajectories. Relatedly, other important contextual factors, such as discrimination, were not available in NHATS and thus cannot be tested. Last, the identified number of trajectory groups was subject to uncertainty due to the data-driven modeling process of the group-based trajectory modeling approach (Ferro & Speechley, 2013). The identified number of trajectory groups was based on model fit and probability of group membership, thus should be interpreted as suggestive.

Despite limitations, this study has several noteworthy strengths. First, we used group-based trajectory modeling to account for non-random attrition. Using methods that assume random missingness is susceptible to biased estimation when the attrition differs by trajectory groups (Haviland et al., 2011). Second, neighborhood characteristics were captured by a combination of self-reported social cohesion and interviewer-observed physical environment, which provides both subjective and objective measures of the neighborhood context. Last, we used a large, nationally representative sample of older African Americans with extended follow-up period, which provides a valuable opportunity to estimate distinct depressive symptom trajectories for this understudied group. It is recommended that future directions on racial minority groups focus on within-group design (Whitfield et al., 2008).

Conclusion

Elevated depressive symptoms among older African Americans may follow three distinct trajectory groups over 9 years. Our findings highlight the potential protective effects of neighborhood social cohesion against the risk of increasing depressive symptoms and the moderating role of gender in the association between neighborhood disadvantage and depressive symptom trajectories. Our findings emphasize the need to promote mental health among older African Americans at the neighborhood-level, especially among men living in disadvantaged neighborhoods. Additionally, this study also highlights the possible individuals-level risk factors of increasing depressive symptoms among older African Americans, including low income levels, cognitive impairment, and prior hospitalization. Enhanced efforts are needed to provide accessible and affordable mental health services among older African Americans who are socioeconomically disadvantaged and have poorer cognitive health. Future research is needed to further clarify the mechanisms underlying the gender difference in the association between neighborhoods and long-term depressive symptoms, as well as how individual-level and community-level factors interplay to influence late-life depressive symptoms among older African Americans.

Supplementary Material

Supp 1

Acknowledgement

This study was presented in an oral paper session at the Annual Conference of the Society for Social Work Research on January 12, 2023, Phoenix, AZ.

Funding

The preparation of this article was supported by the National Institute on Aging of the National Institutes of Health (T32AG000221) (W.Q.), National Institute on Aging of the National Institutes of Health (P30AG015281) and the Michigan Center for Urban African American Aging Research (C.L.E.), and the National Institute on Aging of the National Institutes of Health (P30AG072959) (A.W.N). Support was also received from grant P2CHD042849, awarded to the Population Research Center at The University of Texas at Austin by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (C.L.E.). The National Health and Aging Trends Study (NHATS) is sponsored by the National Institute on Aging of the National Institutes of Health (U01AG032947) and was conducted by the Johns Hopkins University and University of Michigan. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Footnotes

Declaration of Interest Statement

The authors report there are no competing interests to declare.

Contributor Information

Weidi Qin, Population Studies Center, Institute for Social Research, University of Michigan, Ann Arbor, MI.

Christy L. Erving, Department of Sociology, The University of Texas at Austin, Austin, TX.

Ann W. Nguyen, Jack, Joseph and Morton Mandel School of Applied Social Sciences, Case Western Reserve University, Cleveland, OH.

References

  1. Andrews MR, Ceasar J, Tamura K, Langerman SD, Mitchell VM, Collins BS, Baumer Y, Gutierrez Huerta CA, Dey AK, Playford MP, Mehta NN, & Powell-Wiley TM (2021). Neighborhood environment perceptions associate with depression levels and cardiovascular risk among middle-aged and older adults: Data from the Washington, DC cardiovascular health and needs assessment. Aging & Mental Health, 25(11), 2078–2089. 10.1080/13607863.2020.1793898 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Barnett A, Zhang CJP, Johnston JM, & Cerin E (2018). Relationships between the neighborhood environment and depression in older adults: a systematic review and meta-analysis. International Psychogeriatrics, 30(8), 1153–1176. 10.1017/S104161021700271X [DOI] [PubMed] [Google Scholar]
  3. Barr PB (2018). Early neighborhood conditions and trajectories of depressive symptoms across adolescence and into adulthood. Advances in Life Course Research, 35, 57–68. 10.1016/j.alcr.2018.01.005 [DOI] [Google Scholar]
  4. Berke EM, Gottlieb LM, Moudon AV, & Larson EB (2007). Protective Association Between Neighborhood Walkability and Depression in Older Men. Journal of the American Geriatrics Society, 55(4), 526–533. 10.1111/j.1532-5415.2007.01108.x [DOI] [PubMed] [Google Scholar]
  5. Byers AL, Vittinghoff E, Lui L-Y, Hoang T, Blazer DG, Covinsky KE, Ensrud KE, Cauley JA, Hillier TA, Fredman L, & Yaffe K (2012). Twenty-Year Depressive Trajectories Among Older Women. Archives of General Psychiatry, 69(10), 1073–1079. 10.1001/archgenpsychiatry.2012.43 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Campbell-Grossman C, Hudson DB, Kupzyk KA, Brown SE, Hanna KM, & Yates BC (2016). Low-Income, African American, Adolescent Mothers’ Depressive Symptoms, Perceived Stress, and Social Support. Journal of Child and Family Studies, 25(7), 2306–2314. 10.1007/s10826-016-0386-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Centers for Disease Control and Prevention. (2021). Depression is Not a Normal Part of Growing Older. https://www.cdc.gov/aging/depression/index.html
  8. Cho S (2022). Relationships Between Perceived Neighborhood Disorder and Depressive Symptomatology: The Stress Buffering Effects of Social Support Among Older Adults. Social Work in Public Health, 37(1), 45–56. 10.1080/19371918.2021.1974636 [DOI] [PubMed] [Google Scholar]
  9. Choi NG, Kim J, DiNitto DM, & Marti CN (2015). Perceived Social Cohesion, Frequency of Going Out, and Depressive Symptoms in Older Adults: Examination of Longitudinal Relationships. Gerontology and Geriatric Medicine, 1, 2333721415615478. 10.1177/2333721415615478 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Echeverría S, Diez-Roux AV, Shea S, Borrell LN, & Jackson S (2008). Associations of neighborhood problems and neighborhood social cohesion with mental health and health behaviors: The Multi-Ethnic Study of Atherosclerosis. Health & Place, 14(4), 853–865. 10.1016/j.healthplace.2008.01.004 [DOI] [PubMed] [Google Scholar]
  11. Erving CL (2022). Gender, Stress, and Mental Health Among Older African Americans. Annual Review of Gerontology and Geriatrics, Chapter 6, 145–182. 10.1891/0198-8794.41.145 [DOI] [Google Scholar]
  12. Erving CL, & Cobb RJ (2021). Neighborhood Social Group Participation and Depressive Symptoms Among Mid-to-Late Life Black Americans: Does the Association Differ by Ethnicity? Journal of Immigrant and Minority Health, 23(3), 478–486. 10.1007/s10903-020-01070-x [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Ferro MA, & Speechley KN (2013). Stability of latent classes in group-based trajectory modeling of depressive symptoms in mothers of children with epilepsy: an internal validation study using a bootstrapping procedure. Social Psychiatry and Psychiatric Epidemiology, 48(7), 1077–1086. 10.1007/s00127-012-0622-6 [DOI] [PubMed] [Google Scholar]
  14. Firebaugh G, & Acciai F (2016). For blacks in America, the gap in neighborhood poverty has declined faster than segregation. Proceedings of the National Academy of Sciences, 113(47), 13372–13377. 10.1073/pnas.1607220113 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Freedman VA, Kasper JD, Spillman BC, Agree EM, Mor V, Wallace RB, & Wolf DA (2013). Behavioral Adaptation and Late-Life Disability: A New Spectrum for Assessing Public Health Impacts. American Journal of Public Health, 104(2), e88–e94. 10.2105/AJPH.2013.301687 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Gill TM, & Williams CS (2017). Evaluating Distinctions in the Assessment of Late-Life Disability. The Journals of Gerontology: Series A, 72(11), 1538–1546. 10.1093/gerona/glx022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gutman LM, & Sameroff AJ (2004). Continuities in depression from adolescence to young adulthood: Contrasting ecological influences. Development and Psychopathology, 16(4), 967–984. 10.1017/S095457940404009X [DOI] [PubMed] [Google Scholar]
  18. Haigh EAP, Bogucki OE, Sigmon ST, & Blazer DG (2018). Depression Among Older Adults: A 20-Year Update on Five Common Myths and Misconceptions. The American Journal of Geriatric Psychiatry, 26(1), 107–122. 10.1016/j.jagp.2017.06.011 [DOI] [PubMed] [Google Scholar]
  19. Haviland AM, Jones BL, & Nagin DS (2011). Group-based Trajectory Modeling Extended to Account for Nonrandom Participant Attrition. Sociological Methods & Research, 40(2), 367–390. 10.1177/0049124111400041 [DOI] [Google Scholar]
  20. Joint Center for Housing Studies of Harvard University. (2018). Housing America’s older adults 2018: a supplement to the state of the nation’s housing report. www.jchs.harvard.edu/sites/default/files/jchs-housing_americas_older_adults_2014_1.pdf
  21. Jones BL, & Nagin DS (2013). A Note on a Stata Plugin for Estimating Group-based Trajectory Models. Sociological Methods & Research, 42(4), 608–613. 10.1177/0049124113503141 [DOI] [Google Scholar]
  22. Kasper JD, Freedman VA, & Spillman B (2013). Classification of Persons by Dementia Status in the National Health and Aging Trends Study. Technical Paper #5. www.NHATS.org [Google Scholar]
  23. Kessler RC (2012). The Costs of Depression. Psychiatric Clinics, 35(1), 1–14. 10.1016/j.psc.2011.11.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Killeen OJ, Xiang X, Powell D, Reed NS, Deal JA, Swenor BK, & Ehrlich JR (2022). Longitudinal Associations of Self-Reported Visual, Hearing, and Dual Sensory Difficulties With Symptoms of Depression Among Older Adults in the United States. Frontiers in Neuroscience, 16. 10.3389/fnins.2022.786244 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kim D (2008). Blues from the Neighborhood? Neighborhood Characteristics and Depression. Epidemiologic Reviews, 30(1), 101–117. 10.1093/epirev/mxn009 [DOI] [PubMed] [Google Scholar]
  26. Kuchibhatla MN, Fillenbaum GG, Hybels CF, & Blazer DG (2012). Trajectory classes of depressive symptoms in a community sample of older adults. Acta Psychiatrica Scandinavica, 125(6), 492–501. 10.1111/j.1600-0447.2011.01801.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Latham K, & Clarke P (2016). Neighborhood Disorder, Perceived Social Cohesion, and Social Participation Among Older Americans: Findings From the National Health & Aging Trends Study. Journal of Aging and Health, 30(1), 3–26. 10.1177/0898264316665933 [DOI] [PubMed] [Google Scholar]
  28. Latkin CA, & Curry AD (2003). Stressful Neighborhoods and Depression: A Prospective Study of the Impact of Neighborhood Disorder. Journal of Health and Social Behavior, 44(1), 34–44. 10.2307/1519814 [DOI] [PubMed] [Google Scholar]
  29. Lawton MP (1999). Environmental taxonomy: Generalizations from research with older adults. In Friedman S & Wachs T (Eds.), Measuring environment across the life span (pp. 91–124). American Psychological Association. [Google Scholar]
  30. Li C, Friedman B, Conwell Y, & Fiscella K (2007). Validity of the Patient Health Questionnaire 2 (PHQ-2) in Identifying Major Depression in Older People. Journal of the American Geriatrics Society, 55(4), 596–602. 10.1111/j.1532-5415.2007.01103.x [DOI] [PubMed] [Google Scholar]
  31. Lincoln KD, Chatters LM, Taylor RJ, & Jackson JS (2007). Profiles of depressive symptoms among African Americans and Caribbean Blacks. Social Science & Medicine, 65(2), 200–213. 10.1016/j.socscimed.2007.02.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Louie P, Upenieks L, Erving CL, & Thomas Tobin CS (2021). Do Racial Differences in Coping Resources Explain the Black–White Paradox in Mental Health? A Test of Multiple Mechanisms. Journal of Health and Social Behavior, 63(1), 55–70. 10.1177/00221465211041031 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Montagnier D, Dartigues J-F, Rouillon F, Pérès K, Falissard B, & Onen F (2014). Ageing and trajectories of depressive symptoms in community-dwelling men and women. International Journal of Geriatric Psychiatry, 29(7), 720–729. 10.1002/gps.4054 [DOI] [PubMed] [Google Scholar]
  34. Montaquila J, Freedman V, Edwards B, & Kasper JD (2012). National Health and Aging Trends Study Round 1 Sample Design and Selection. https://www.nhats.org [Google Scholar]
  35. Musliner KL, Munk-Olsen T, Eaton WW, & Zandi PP (2016). Heterogeneity in long-term trajectories of depressive symptoms: Patterns, predictors and outcomes. Journal of Affective Disorders, 192, 199–211. 10.1016/j.jad.2015.12.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Nagin DS (2010). Group-Based Trajectory Modeling: An Overview BT - Handbook of Quantitative Criminology. In Piquero AR & Weisburd D (Eds.), Handbook of quantitative criminology (pp. 53–67). Springer; New York. 10.1007/978-0-387-77650-7_4 [DOI] [Google Scholar]
  37. Paczkowski MM, & Galea S (2010). Sociodemographic characteristics of the neighborhood and depressive symptoms. Current Opinion in Psychiatry, 23(4). https://journals.lww.com/co-psychiatry/Fulltext/2010/07000/Sociodemographic_characteristics_of_the.8.aspx [DOI] [PubMed] [Google Scholar]
  38. Park M (2017). Impact of social capital on depression trajectories of older women in Korea. Aging & Mental Health, 21(4), 354–361. 10.1080/13607863.2015.1088511 [DOI] [PubMed] [Google Scholar]
  39. Pratt LA, & Brody DJ (2014). Depression in the US household population, 2009–2012 (Issue NCHS data brief, no 172). Hyattsville, MD: National Center for Health Statistics. [PubMed] [Google Scholar]
  40. Qin W, Clarke PJ, & Ehrlich JR (2022). Self-Reported Visual Difficulty and Daily Activity Limitations: The Moderating Role of Neighborhood Characteristics. The Gerontologist, gnac 143. 10.1093/geront/gnac143 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Repetto PB, Caldwell CH, & Zimmerman MA (2004). Trajectories of depressive symptoms among high risk African-American adolescents. Journal of Adolescent Health, 35(6), 468–477. 10.1016/j.jadohealth.2003.12.007 [DOI] [PubMed] [Google Scholar]
  42. Ruiz M, Scholes S, & Bobak M (2018). Perceived neighbourhood social cohesion and depressive symptom trajectories in older adults: a 12-year prospective cohort study. Social Psychiatry and Psychiatric Epidemiology, 53(10), 1081–1090. 10.1007/s00127-018-1548-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Scheidt RJ, & Norris-Baker C (2003). The general ecological model revisited: Evolution, current status, and continuing challenges. In Wahl HW, Scheidt RJ, Windley PG, & Schaie KW (Eds.), Annual Review of Gerontology and Geriatrics; (pp. 34–58). [Google Scholar]
  44. Skolarus LE, Sánchez BN, Morgenstern LB, Garcia NM, Smith MA, Brown DL, & Lisabeth LD (2010). Validity of Proxies and Correction for Proxy Use When Evaluating Social Determinants of Health in Stroke Patients. Stroke, 41(3), 510–515. 10.1161/STROKEAHA.109.571703 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. StataCorp. (2021). Stata Statistical Software: Release 17. College Station, TX: StataCorp LLC. [Google Scholar]
  46. Thomas Tobin CS, Erving CL, Hargrove TW, & Satcher LA (2022). Is the Black-White mental health paradox consistent across age, gender, and psychiatric disorders? Aging & Mental Health, 26(1), 196–204. 10.1080/13607863.2020.1855627 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Whitfield KE, Allaire JC, Belue R, & Edwards CL (2008). Are Comparisons the Answer to Understanding Behavioral Aspects of Aging in Racial and Ethnic Groups? The Journals of Gerontology: Series B, 63(5), P301–P308. 10.1093/geronb/63.5.P301 [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Williams DR, & Jackson PB (2005). Social Sources Of Racial Disparities In Health. Health Affairs, 24(2), 325–334. 10.1377/hlthaff.24.2.325 [DOI] [PubMed] [Google Scholar]
  49. Wolinsky FD, Jones MP, & Wehby GL (2012). Gathering data from older adults via proxy respondents: research challenges. Journal of Comparative Effectiveness Research, 1(6), 467–470. 10.2217/cer.12.54 [DOI] [PubMed] [Google Scholar]
  50. Wong R, & Wang Y (2022). Role of Neighborhood Physical Disorder and Social Cohesion on Racial and Ethnic Disparities in Dementia Risk. Journal of Aging and Health, 08982643221101352. 10.1177/08982643221101352 [DOI] [PubMed] [Google Scholar]
  51. Woodward AT, Taylor RJ, Abelson JM, & Matusko N (2013). Major Depressive Disorder Among Older African Americans, Caribbean Blacks, and Non-Hispanic Whites: Secondary Analysis of the National Survey of American Life. Depression and Anxiety, 30(6), 589–597. 10.1002/da.22041 [DOI] [PubMed] [Google Scholar]
  52. World Health Organization. (2021). Depression. www.who.int/news-room/fact-sheets/detail/depression
  53. Xiang X (2019). Seven-Year Trajectories of Depressive Symptoms and Their Predictors Among Older Americans. Journal of Aging and Health, 32(7–8), 795–806. 10.1177/0898264319852835 [DOI] [PubMed] [Google Scholar]
  54. Xiang X, & Cheng J (2019). Trajectories of major depression in middle-aged and older adults: A population-based study. International Journal of Geriatric Psychiatry, 34, 1506–1514. 10.1002/gps.5161 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp 1
NIHMS1917993-supplement-Supp_1.docx (32.4KB, docx)

RESOURCES