Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jun 1.
Published in final edited form as: Am J Geriatr Psychiatry. 2021 Nov 27;30(6):703–716. doi: 10.1016/j.jagp.2021.11.010

Geographic region, racial/ethnic disparities and late-life depression: results from a large US cohort of older adults

Chirag M Vyas 1, Charles F Reynolds III 2, Macarius Donneyong 3, David Mischoulon 1, Grace Chang 4, Nancy R Cook 5,6, JoAnn E Manson 5,6,7, Olivia I Okereke 1,6,7
PMCID: PMC9106835  NIHMSID: NIHMS1758337  PMID: 34969584

Abstract

Objectives:

To determine associations between geographic region and late-life depression (LLD) severity, item-level symptom burden, and treatment; to evaluate whether racial/ethnic disparities in LLD, previously observed in the overall sample, vary by region.

Methods:

We included 25,502 VITAL (Vitamin D and Omega-3 Trial) participants and administered the Patient Health Questionnaire-8 for depressive symptoms; participants also reported medication and/or counseling care for depression. Multivariable regression analyses were performed.

Results:

Despite overall lower LLD severity and item-level symptom burden in the Midwest vs. Northeast, higher LLD severity and item-level burden were observed among minorities, especially Black and Hispanic adults, compared to non-Hispanic whites in this region. Racial/ethnic disparities in item-level symptoms (e.g., anhedonia, sadness, psychomotor changes) varied by region. There were no significant differences in depression care by region; furthermore, regional variation was not observed in racial disparities in care: e.g., among those with clinician/physician-diagnosed depression, Blacks vs. non-Hispanic whites had >50% lower odds of treatment in all regions.

Conclusion:

LLD varied by geographic region. Furthermore, magnitudes of racial/ethnic disparities in LLD severity and item-level symptom burden, but not depression care, differed by region.

Keywords: Geography, Depression, Race/ethnicity

OBJECTIVES

Racial/ethnic disparities in late-life depression (LLD) prevalence and care have been observed in the United States (US).1,2 Recently, we reported racial/ethnic disparities in LLD severity, item-level symptom burden, and care.3 Geographic region is an understudied depression risk factor; further, it could be an explanatory factor in racial/ethnic disparities in LLD.4–8

US Geographic regions have diverse racial/ethnic distributions, and region may serve as a proxy for shared social and health determinants, such as socio-economic, neighborhood and community factors relevant to depression risk.9 Prior literature has suggested significant differences in depression prevalence rates across the US after accounting for sociodemographic characteristics.10 However, a comprehensive analysis, incorporating numerous covariates, of the role of geographic region in LLD is lacking; yet, data from such analyses may have substantial implications for LLD prevention.11

We previously reported that older racial/ethnic minorities, especially Blacks and Hispanics, had significantly higher LLD severity and item-level symptom burden than non-Hispanic whites after adjusting for numerous depression risk factors, in a large national cohort (N>25,000); yet, Black participants had significantly lower likelihood of depression care compared to non-Hispanic whites, even when concurrently reporting clinician-diagnosed depression and comparable LLD severity levels.3 Furthermore, racial/ethnic differences in LLD by geographic region were noted. In a separate cohort of older women (N=29,771), we observed an interaction between race/ethnicity and geographic birth region for LLD incidence: being born in the South was associated with lower LLD risk among Blacks but higher risk among non-Hispanic whites.12 Racial/ethnic disparities in mental health treatment in all regions except the Northeast have also been observed.8 However, to date, racial/ethnic disparities in LLD severity, item-level symptoms, and treatment by region have not been examined in a large, well-characterized, and diverse cohort. Previous work has emphasized the importance of de-convoluting the role of geographic region in associations between race/ethnicity and brain outcomes (e.g., stroke);13,14 however, data are limited on this issue with regards to LLD.

Therefore, we examined the role of geographic region in LLD severity, symptom burden, and care among over 25,000 adults. Further, we examined whether associations between race/ethnicity and LLD outcomes reported previously in this sample additionally varied by region.3

METHODS

Sample

Participants were members of the VITAL-DEP (VITamin D and OmegA-3 TriaL-Depression Endpoint Prevention) ancillary study to VITAL.15,16 VITAL included 25,871 men and women, aged 50+ and 55+ years, respectively, randomized in 2x2 factorial design to vitamin D3 and/or marine omega-3 fatty acids; study details appear elsewhere.15,16 VITAL recruited participants throughout the US, with attention to achieving comparable participation in each of four US regions. After excluding participants with missing data on depression (n=368) or region (n=1), we included 25,502 participants for analyses (Figure S1). Informed consent was obtained, and the study was approved by the Institutional Review Board at Brigham and Women’s Hospital. Participation in the study is voluntary.

Assessment of geographic regions

Geographic regions of residence (Northeast, Midwest, Southeast, and West [including the far West, Southwest, and mountain regions]) were self-reported on the baseline questionnaire. See Table S1 for the list of states by US region.

Assessment of race/ethnicity

Race/ethnicity (non-Hispanic white, Black, Hispanic, Asian, and Other race/ethnicity [i.e., Native American/Alaska Native, Native Hawaiian or other Pacific Islander, and those reporting multiple race, unknown race and/or unknown ethnicity]) was self-reported on the baseline questionnaire.

Assessment and measures of depression

Depression status was characterized by presence of clinically relevant depressive symptoms, diagnosis, and/or treatment, using baseline questionnaire data. To ascertain symptoms, we used the Patient Health Questionnaire-8 (PHQ-8),17,18 which is validated for clinical depression (PHQ-8≥10)17 and for use among diverse samples of older adults.19 Depression severity was defined by total PHQ-8 score. Item-level symptom burden was defined as reporting the given symptom “more than half the days” or “nearly every day” on the PHQ-8. Depression care was determined by participant self-report of clinician-diagnosed depression and use of antidepressant medication and/or depression counseling.

Statistical analyses

Participants’ characteristics were compared by region. Distributions of depression-related variables among racial/ethnic groups were contrasted by region. First, as PHQ-8 scores were non-normally distributed, the Kruskal-Wallis test was used to examine differences in PHQ-8 scores by geographic region. Then, for the main analyses, we used multivariable zero-inflated negative binomial (ZINB) regression models, as previously,3 to examine the association between geographic region and LLD severity. The NB portion indicated a rate ratio (RR) for a one-point increment in PHQ-8 score; higher counts indicated higher depression severity; the ZI portion indicated the likelihood of “excess zeros” [i.e., 0 vs. ≥1 points on the PHQ-8]. We also used multivariable ZINB models to estimate racial/ethnic differences in LLD severity by region. Additionally, we computed the intraclass correlation coefficient (ICC) to identify potential clustering effects within regions using generalized linear mixed models (GLMM). We used multivariable logistic regression to analyze associations of geographic region with item-level symptom burden and depression care, and racial/ethnic differences therein. Depression care was evaluated among participants who endorsed clinically significant depressive symptoms (PHQ-8≥10);17 a separate analysis was performed among participants who reported clinician-diagnosed depression.

All regression models were sequentially adjusted for socio-demographic factors (age, sex, education, income), and then for lifestyle/behavioral factors [body mass index (BMI), physical activity (PA), alcohol and smoking use], and finally for medical comorbidities (hypertension, diabetes, high cholesterol). In analyses of geographic region and LLD outcomes, a two-tailed test was used for statistical significance at alpha=0.05. In analyses of racial/ethnic differences in LLD outcomes by region, we performed multiple hypothesis testing at false discovery rate (FDR)<0.05.20 Analyses were performed in SAS 9.4 (SAS Institute, Cary, NC).

RESULTS

Sample

Table 1 shows descriptive characteristics for the entire sample and by region. The mean (SD) age of study participants was 67.1 (7.1) years; 50.5% were women and 45.3% had post-graduate education. Of the four US regions, the Southeast and Northeast had the largest proportions of participants. The largest proportion of non-Hispanic whites was in the Northeast, the largest proportion of Blacks was in the Southeast, and the largest proportions of Hispanics and Asians were in the West, compared to other regions. Participants in the Midwest and Southeast had lower educational attainment and annual household income than those in other regions. There were also regional differences in distributions of lifestyle/behavioral factors: higher BMI, lower PA, and higher current smoking were reported in Midwestern and Southeastern regions, while daily alcohol consumption was more prevalent in Northeastern and Western regions. Prevalence of hypertension, diabetes, and high cholesterol was slightly higher in the Southeast than in other regions.

Table 1.

Participant Characteristics by Geographic Region.*

Participant characteristics Total samplea (n=25,502) Geographic region
Northeast (n=7066) Southeast (n=7133) Midwest (n=5471) Westb (n=5832)
N % N % N % N % N %
Age (Mean±SD), in years 67.1±7.1 66.6±7.2 66.3±7.2 67.4±7.0 68.5±6.6
Age categories, in years
50–64 9735 38.2 2953 41.8 3105 43.5 2033 37.2 1644 28.2
65–74 12,516 49.1 3285 46.5 3238 45.4 2696 49.3 3297 56.5
≥ 75 3251 12.8 828 11.7 790 11.1 742 13.6 891 15.3
Gender
Men 12,614 49.5 3699 52.4 3532 49.5 2541 46.4 2842 48.7
Women 12,888 50.5 3367 47.7 3601 50.5 2930 53.6 2990 51.3
Race/ethnicity          
Non-Hispanic white 17,828 69.9 5134 72.7 4689 65.7 3861 70.6 4144 71.1
Black 5004 19.6 1355 19.2 1773 24.9 1212 22.2 664 11.4
Hispanic 1001 3.9 192 2.7 189 2.7 87 1.6 533 9.1
Asian 377 1.5 95 1.3 50 0.7 42 0.8 190 3.3
Otherc 1292 5.1 290 4.1 432 6.1 269 4.9 301 5.2
Education        
Did not complete high school 346 1.4 95 1.4 136 1.9 57 1.0 58 1.0
High school diploma or GED 2890 11.4 747 10.6 1012 14.2 618 11.3 513 8.8
Attended or graduated from college 10,697 42.0 2686 38.1 3060 43.0 2509 46.0 2442 41.9
Post-graduate 11,516 45.3 3527 50.0 2905 40.8 2273 41.7 2811 48.3
Income          
<$15,000 1445 6.3 320 5.1 523 8.1 349 7.1 253 4.8
$15,000 – 29,999 2875 12.5 639 10.1 942 14.5 684 14.0 610 11.5
$30,000 – 49,999 4080 17.8 934 14.8 1223 18.8 968 19.8 955 18.1
$50,000 – 69,999 3763 16.4 964 15.3 1093 16.8 844 17.2 862 16.3
$70,000 – 89,999 2919 12.7 786 12.5 759 11.7 661 13.5 713 13.5
$90,000 – 120,000 3736 16.3 1145 18.2 970 14.9 724 14.8 897 17.0
Over $120,000 4174 18.2 1522 24.1 990 15.2 667 13.6 995 18.8
BMI (Mean±SD), kg/m2 28.1±5.7 27.8±5.4 28.5±6.0 28.7±6.1 27.5±5.3
Total physical activity (MET-hours/week)d, median (IQR) 15.5 (4.6–31.7) 17.4 (5.6–33.6) 13.3 (3.5–29.7) 14.5 (4.4–29.9) 17.0 (5.9–34.3)
Alcohol use
Never 7880 31.4 1761 25.7 2538 36.0 1801 33.3 1780 30.8
Monthly 1878 7.5 491 7.2 561 8.0 413 7.6 413 7.2
Weekly 8779 35.0 2608 38.0 2269 32.2 1950 36.1 1952 33.8
Daily 6555 26.1 2006 29.2 1680 23.8 1242 23.0 1627 28.2
Cigarette smoking status          
Never 13,071 51.7 3509 50.1 3618 51.2 2776 51.1 3168 54.8
Past 10,384 41.1 3041 43.5 2836 40.2 2197 40.5 2310 39.9
Current 1820 7.2 449 6.4 609 8.6 457 8.4 305 5.3
Hypertensione          
Yes 13,142 51.8 3515 50.0 3891 54.9 2826 51.9 2910 50.2
No 12,223 48.2 3517 50.0 3196 45.1 2617 48.1 2893 49.9
Diabetesf          
Yes 3495 13.7 877 12.4 1075 15.1 806 14.8 737 12.7
No 21,966 86.3 6179 87.6 6043 84.9 4654 85.2 5090 87.4
High cholesterol        
Yes 9461 37.5 2538 36.3 2713 38.5 2053 37.9 2157 37.4
No 15,757 62.5 4459 63.7 4327 61.5 3361 62.1 3610 62.6
PHQ-8 score,g median (IQR) 1.0 (0.0–2.0) 1.0 (0.0–2.0) 1.0 (0.0–3.0) 1.0 (0.0–2.0) 1.0 (0.0–2.0)
PHQ-8 severity
None (0 points) 11,888 46.6 3329 47.1 3230 45.3 2600 47.5 2729 46.8
Minimal (1–4 points) 10,670 41.8 2951 41.8 2934 41.1 2291 41.9 2494 42.8
Mild (5–9 points) 2189 8.6 581 8.2 704 9.9 448 8.2 456 7.8
Moderate (10–14 points) 504 2.0 142 2.0 169 2.4 86 1.6 107 1.8
Moderate–to-severe (15–19 points) 183 0.7 52 0.7 67 0.9 34 0.6 30 0.5
Severe (20+ points) 68 0.3 11 0.2 29 0.4 12 0.2 16 0.3
SSRI use          
Yes 1603 6.4 433 6.2 525 7.5 319 5.9 326 5.7
No 23,536 93.6 6530 93.8 6505 92.5 5072 94.1 5429 94.3
Diagnosed with depression in past 2 years          
Yes 4846 20.7 1393 21.6 1359 20.7 997 19.8 1097 20.5
No 18,578 79.3 5072 78.5 5204 79.3 4038 80.2 4264 79.5
→If diagnosed, use of Medication/Counseling
Yes 2626 54.9 779 56.6 765 56.9 528 53.5 554 51.4
No 2158 45.1 597 43.4 579 43.1 459 46.5 523 48.6
Open in a new tab

Abbreviations: GED, general education diploma; BMI, body mass index; PHQ8, patient health questionnnaire-8; SSRI, selective serotonin reuptake inhibitor; SD, standard deviation; IQR, inter-quartile range

*

Figures for percentages may not add to 100.0 due to rounding.

a

For categorical variables, this column contains the number of participants in category/% of non-missing responses. For continuous variables, this column contains mean (standard deviation), median (interquartile range: lower quartile-upper quartile) for noncontinuous variables, and for categorical variables frequency percentages for non-missing responses.

b

West region includes far west, southwest, and mountain regions

c

Other race/ethnicity includes Native American/Alaska Native, Native Hawaiian or other Pacific Islander, multiple race or unknown race and/or unknown ethnicity.

d

Leisure-time physical activities: walking or hiking; jogging; running; bicycling; aerobic exercise/aerobic dance/exercise machines; lower intensity exercise/yoga/stretching/toning; tennis/squash/ racquetball; lap swimming; weightlifting/strength training; other exercises

e

Hypertension: Ever diagnosed with high blood pressure or ever use of anti-hypertensive medication

f

Diabetes: Ever diagnosed with diabetes or current use of anti-diabetic medication

g

Kruskal-Wallis test of overall difference in PHQ-8 scores by geographic region: Chi-square=25.90, df=3, p<0.0001.

Regarding depression variables, there was a statistically significant overall difference in PHQ-8 scores by geographic region (Chi-square=25.90, df=3, p<0.0001). Participants in the Midwest had a lower prevalence of PHQ-8≥10 than those in other regions. Depression care was less prevalent in the Midwest than in other regions. Racial/ethnic differences in distributions of depression characteristics by region were noted (Figure S2).

Geographic region and LLD severity

Participants in the Midwest, but not other regions, had significantly lower depression severity than those in the Northeast [RR (95% CI): 0.88 (0.83–0.94)] (Table 2: NB portion). No significant regional differences were detected in the odds of excess zeroes in the PHQ-8 score (Table 2: ZI portion). Complete output from the multivariable ZINB model is shown in Tables S2A–S2B; details regarding the rationale for the ZINB approach are provided elsewhere (Supplementary Methods; Figure S3). Of note, there was no variation in LLD severity by region in GLMM (ICC=0.23%); thus, there were negligible clustering effects in regions for LLD severity.

Table 2.

Association between Geographic Region and Late-life Depression Severity, and Racial/ethnic Differences in Severity of Late-life Depression within Geographic Region*

Geographic region Comparison: Region vs. LLD severitya Comparison: Race/ethnicity vs. LLD severity within each geographic regionc
Non-Hispanic white Black Hispanic Asian Other, multiple or unspecified race/ethnicity
RR/OR (95% CI) RR/OR RR/OR (95% CI) RR/OR (95% CI) RR/OR (95% CI) RR/OR (95% CI)
Negative Binomial (NB) portion (RRs are shown)d
Northeast 1.00 (Ref) 1.00 (Ref) 1.18 (1.06 – 1.32) 1.10 (0.87 – 1.41) 1.03 (0.70 – 1.52) 1.30 (1.06 – 1.60)
Southeast 1.04 (0.98 – 1.10) 1.00 (Ref) 0.97 (0.88 – 1.07) 1.24 (0.97 – 1.58) 1.41 (0.73 – 2.73) 1.14 (0.98 – 1.34)
Midwest 0.88 (0.83 – 0.94)b 1.00 (Ref) 1.23 (1.09 – 1.39) 1.99 (1.37 – 2.89) 1.43 (0.83 – 2.45) 1.23 (0.99 – 1.53)
West 0.97 (0.92 – 1.03) 1.00 (Ref) 1.14 (0.98 – 1.31) 1 .21 (1.03 – 1.43) 1.11 (0.84 – 1.46) 0.99 (0.83 – 1.18)
Zero Inflation (ZI) portion (ORs are shown)d
Northeast 1.00 (Ref) 1.00 (Ref) 2.27 (1.29 – 4.01) 1.23 (0.40 – 3.78) 1.68 (0.35 – 8.07) 1.94 (0.80 – 4.69)
Southeast 1.26 (0.98 – 1.63) 1.00 (Ref) 1.88 (1.34 – 2.62) 1.44 (0.70 – 3.00) 6.52 (2.52 – 16.85) 1.06 (0.60 – 1.89)
Midwest 1.10 (0.82 – 1.48) 1.00 (Ref) 2.90 (1.61 – 5.22) 4.59 (1.94 – 10.90) 3.02 (0.76 – 11.96) 2.31 (1.11 – 4.83)
West 0.84 (0.62 – 1.14) 1.00 (Ref) 2.06 (0.87 – 4.91) 3.63 (1.52 – 8.64) 2.43 (0.68 – 8.66) --
Open in a new tab

Abbreviations: ZINB, zero-inflated negative binomial; RR, rate ratio; OR, odds ratio; CI, confidence interval

Variable definitions are provided in the footnote of Table 1.

a

The northeast region was used as a reference group for the association of geographic region and LLD severity.

b

Wald chi-square test, df=1, p-value <0.05.

c

Non-Hispanic white participants were reference group for patterns of racial/ethnic differences in the severity of LLD within a geographic region. Estimates that remained significant after adjusting for multiple hypothesis testing using a Wald chi-square test (df=1) at false discovery rate<0.05 are highlighted in bold. Two separate FDR corrections performed for NB and ZI portions of ZINB (i.e., 16 tests were included in each comparison).

d

The estimates shown were adjusted for demographic, lifestyle/behavioral, and health related factors. To avoid undefined physical activity estimates when using a missing indicator for physical activity, we imputed the median value for the small percentage (<1%) of participants who were missing information on physical activity.

*

Results from the negative binomial (NB) portion of zero-inflated negative binomial (ZINB) regression show rate ratios (RRs) and 95% confidence intervals (CIs), which reflect percent differences in total depression severity on the PHQ-8 in total sample and different racial/ethnic groups within each geographic region. Results from the zero-inflated (ZI) portion of ZINB regression show odds ratios (OR) and 95% confidence intervals (CIs), which reflect likelihood of “excess zeros” i.e., all zeros (0 points) vs. ≥1 point on the PHQ-8 among different racial/ethnic groups within each geographic region. The ZI portion is not readily clinically interpretable by itself in this context due to a lack of distinction of “true” vs. “excess” zeroes using the PHQ-8, and the latent process for zeroes is not relevant for grading depression symptom severity. Predictors for ZI and NB portions of the model cannot be assumed to be the same. However, it can adjust the estimated probabilities of higher scores. For completeness, estimates for all predictors in the fully-adjusted model are shown for both the ZI and NB portions.

Race/ethnicity and LLD severity by region

In the entire sample Black, Hispanic, and Other race/ethnicity participants had 10%, 23%, and 14% higher LLD severity, respectively, compared to non-Hispanic whites, as reported previously.3 Adding to these findings, we found evidence of geographic variation in these observed racial/ethnic differences in LLD severity (Table 2): after FDR correction, in the Northeast, Black and Other race/ethnicity participants had, respectively, 18% and 30% higher LLD severity compared to non-Hispanic whites; in the Midwest, Blacks and Hispanics had respectively, 23% and 99% higher LLD severity than non-Hispanic whites. Racial/ethnic differences in depression severity levels in the Southeast and West were not observed. We included 16 tests (i.e., 4 racial/ethnic subgroups in all 4 regions) for FDR correction.

Regarding ZI results of ZINB, Blacks, Hispanics, and Asians had 2–2.5 times higher odds of excess zeros compared to non-Hispanic whites in the entire sample, as previously reported;3 in the current analysis (Table 2), we observed higher odds of excess zeroes among minorities compared to non-Hispanic whites, with some evidence of variation by geographic region. We included 16 tests for FDR correction. Detailed comparisons between observed vs. adjusted ZINB proportions of racial/ethnic groups among PHQ-8 categories by region have been provided (Figure S4); these suggest that included model covariates help to explain part of racial/ethnic disparity.

Geographic region and item-level symptom burden

Participants in the Midwest had a significantly lower burden of several item-level depression symptoms (e.g., 20–30% lower multivariable-adjusted odds of sadness, sleep, guilt, and concentration problems) than did those in the Northeast (Table 3). We observed significantly higher multivariable-adjusted odds of energy problems in the Southeast compared to the Northeast.

Table 3.

Adjusted Odds Ratio (OR) of Item-level Symptom Burden by Geographic Region and Patterns of Racial/ethnic Differences in the Item-level Depressive Symptoms within Geographic Region.

Item-level symptom US Geographic region Comparison: Region vs. symptom burdena Comparison: Race/ethnicity vs. symptom burden within each geographic regionc
Non-Hispanic white Black Hispanic Asian Other
Adjusted OR (95% CI)d
Anhedonia Northeast 1.00 (ref) 1.00 (ref) 2.48 (1.73 – 3.57) 1.73 (0.83 – 3.61) --e 1.99 (1.05 – 3.78)
Southeast 1.09 (0.89 – 1.32) 1.00 (ref) 1.11 (0.82 – 1.52) 1.95 (0.98 – 3.87) 0.96 (0.13 – 7.36) 1.35 (0.83 – 2.19)
Midwest 0.87 (0.69 – 1.09) 1.00 (ref) 2.02 (1.32 – 3.08) 3.50 (1.32 – 9.32) 6.60 (1.85 – 23.51) 1.64 (0.78 – 3.47)
West 1.12 (0.90 – 1.40) 1.00 (ref) 2.49 (1.61 – 3.85) 2.05 (1.25 – 3.35) 2.83 (1.40 – 5.74) 0.95 (0.41 – 2.23)
Sadness Northeast 1.00 (ref) 1.00 (ref) 1.86 (1.29 – 2.70) 2.04 (1.03 – 4.04) --e 1.85 (0.98 – 3.51)
Southeast 0.87 (0.70 – 1.06) 1.00 (ref) 0.95 (0.67 – 1.34) 1.72 (0.77 – 3.84) 1.08 (0.14 – 8.33) 1.92 (1.18 – 3.10)
Midwest 0.69 (0.54 – 0.88)b 1.00 (ref) 1.62 (1.01 – 2.60) 7.21 (3.15 – 16.48) 5.02 (1.12 – 22.59) 0.85 (0.30 – 2.46)
West 0.89 (0.71 – 1.13) 1.00 (ref) 1.15 (0.69 – 1.92) 1.75 (1.04 – 2.93) 1.43 (0.56 – 3.68) 0.74 (0.29 – 1.87)
Sleep Northeast 1.00 (ref) 1.00 (ref) 1.13 (0.91 – 1.40) 1.09 (0.68 – 1.75) 0.65 (0.26 – 1.64) 1.04 (0.69 – 1.55)
Southeast 1.02 (0.91 – 1.14) 1.00 (ref) 0.86 (0.70 – 1.05) 1.45 (0.93 – 2.26) 0.46 (0.11 – 1.93) 1.23 (0.91 – 1.67)
Midwest 0.80 (0.71 – 0.91)b 1.00 (ref) 0.94 (0.73 – 1.21) 2.01 (1.06 – 3.81) 1.67 (0.58 – 4.82) 0.74 (0.44 – 1.23)
West 1.02 (0.90 – 1.15) 1.00 (ref) 1.07 (0.81 – 1.41) 1.01 (0.74 – 1.37) 0.82 (0.47 – 1.45) 1.00 (0.67 – 1.49)
Energy Northeast 1.00 (ref) 1.00 (ref) 0.99 (0.76 – 1.29) 0.85 (0.46 – 1.56) 1.24 (0.49 – 3.15) 1.14 (0.72 – 1.83)
Southeast 1.27 (1.11 – 1.45)b 1.00 (ref) 0.85 (0.68 – 1.05) 1.24 (0.72 – 2.13) 0.71 (0.16 – 3.19) 1.17 (0.83 – 1.65)
Midwest 0.89 (0.76 – 1.04) 1.00 (ref) 1.18 (0.88 – 1.57) 2.15 (1.02 – 4.49) 1.43 (0.33 – 6.17) 1.39 (0.85 – 2.27)
West 1.10 (0.94 – 1.28) 1.00 (ref) 0.91 (0.65 – 1.27) 0.94 (0.64 – 1.38) 1.34 (0.75 – 2.41) 1.03 (0.64 – 1.67)
Appetite Northeast 1.00 (ref) 1.00 (ref) 1.23 (0.90 – 1.68) 1.70 (0.93 – 3.11) 1.40 (0.43 – 4.57) 1.21 (0.68 – 2.14)
Southeast 0.97 (0.82 – 1.14) 1.00 (ref) 0.70 (0.53 – 0.92) 0.63 (0.25 – 1.56) --e 1.19 (0.77 – 1.82)
Midwest 0.83 (0.69 – 1.00) 1.00 (ref) 1.23 (0.87 – 1.73) 3.69 (1.68 – 8.07) 1.31 (0.18 – 9.81) 0.97 (0.49 – 1.94)
West 1.07 (0.89 – 1.29) 1.00 (ref) 1.09 (0.74 – 1.61) 1.02 (0.64 – 1.64) 1.01 (0.43 – 2.37) 0.98 (0.54 – 1.78)
Guilt Northeast 1.00 (ref) 1.00 (ref) 1.27 (0.87 – 1.84) 1.27 (0.61 – 2.68) 1.36 (0.32 – 5.74) 0.91 (0.41 – 2.03)
Southeast 0.88 (0.71 – 1.08) 1.00 (ref) 1.17 (0.83 – 1.65) 2.37 (1.11 – 5.05) 2.44 (0.55 – 10.94) 1.75 (1.04 – 2.96)
Midwest 0.67 (0.53 – 0.86)b 1.00 (ref) 1.32 (0.81 – 2.15) 4.25 (1.57 – 11.50) 5.94 (1.29 – 27.24) 1.64 (0.70 – 3.81)
West 1.00 (0.79 – 1.27) 1.00 (ref) 1.23 (0.76 – 1.98) 1.41 (0.81 – 2.44) 0.56 (0.13 – 2.33) 0.89 (0.38 – 2.10)
Concentration Northeast 1.00 (ref) 1.00 (ref) 1.65 (1.08 – 2.51) 1.99 (0.94 – 4.23) 2.79 (0.84 – 9.31) 1.33 (0.59 – 3.01)
Southeast 0.93 (0.73 – 1.17) 1.00 (ref) 1.08 (0.75 – 1.57) 0.76 (0.23 – 2.48) 1.40 (0.18 – 10.66) 0.94 (0.49 – 1.83)
Midwest 0.66 (0.50 – 0.87)b 1.00 (ref) 2.01 (1.13 – 3.56) 5.13 (1.71 – 15.36) 3.55 (0.45 – 27.94) 2.70 (1.13 – 6.43)
West 0.87 (0.66 – 1.15) 1.00 (ref) 1.64 (0.94 – 2.88) 0.74 (0.32 – 1.69) 2.26 (0.86 – 5.97) 0.66 (0.20 – 2.17)
Psychomotor Northeast 1.00 (ref) 1.00 (ref) 1.54 (0.81 – 2.91) 1.23 (0.35 – 4.42) 1.43 (0.18 – 11.32) 1.33 (0.45 – 3.99)
Southeast 1.31 (0.95 – 1.80) 1.00 (ref) 1.54 (0.95 – 2.50) 4.72 (2.13 – 10.44) --e 1.69 (0.80 – 3.56)
Midwest 0.77 (0.52 – 1.14) 1.00 (ref) 3.36 (1.47 – 7.71) 7.87 (1.64 – 37.78) 10.94 (1.31 – 91.58) 4.23 (1.28 – 14.00)
West 1.17 (0.80 – 1.70) 1.00 (ref) 1.65 (0.82 – 3.34) 0.97 (0.38 – 2.45) 0.64 (0.09 – 4.84) 0.66 (0.15 – 2.82)
Open in a new tab

Abbreviation: OR, odds ratio; CI, confidence interval

Variable definitions are provided in the footnotes to Table 1.

a

The northeast region was used as a reference group for the association of geographic region and item-level symptom burden.

b

Wald chi-square test, df=1, p-value <0.05.

c

Non-Hispanic white participants were reference group for patterns of racial/ethnic differences in the item-level symptom burden. within a geographic region. Estimates that remained significant after multiple hypothesis testing using a Wald chi-square test (df=1) at false discovery rate (FDR)<0.05 are highlighted in bold. FDR correction was performed by item-level symptoms; 8 different FDR corrections were made. In each item-level specific FDR correction, 16 tests were included (4 racial/ethnic subgroup comparisons in all 4 regions).

d

The ORs shown were adjusted for demographic, lifestyle/behavioral, and health-related factors. For both analyses, to avoid undefined physical activity estimates when using a missing indicator for physical activity, we imputed the median value for the small percentage (<1%) of participants who were missing information on physical activity. In a stratified analysis of race/ethnicity and item-level symptom burden by geographic region, we created separate datasets for each geographic region. In these datasets, to avoid quasi-separation issues/undefined estimates in the adjusted models, we imputed missing category to the reference category. a) For the Northeast region, the missing education category was imputed to the reference (‘did not complete high school’) category and participants with missing information on diabetes were imputed to reference (‘no’) category. b) For the Southeast region, the missing education category was imputed to the reference (‘did not complete high school’) category and participants with missing information on smoking and diabetes were imputed to reference (‘no’) category. c) For the Midwest region, the missing education category was imputed to the reference (‘did not complete high school’) category; participants with missing information on smoking, hypertension and diabetes were imputed to reference (‘no’) category. d) For the West region, the missing education category was imputed to reference (‘did not complete high school’), while participants with missing information on hypertension, diabetes and high cholesterol were imputed to reference (‘no’) category.

e

Due to a few participants in the racial/ethnic group, we were unable to calculate the odds ratios.

Race/ethnicity and item-level symptom burden by region

As previously seen in this sample, minorities had overall 1.5–2-fold higher multivariable odds for most items except neurovegetative symptoms (sleep, energy, appetite, psychomotor changes); higher burden from sleep problems and guilt was observed only among Hispanics after adjusting for confounders.3 Table 3 shows variation by region in associations between race/ethnicity and symptom burden, after FDR correction. Black, Hispanic, and Asian participants had ≥2-fold multivariable-adjusted odds of anhedonia, compared to non-Hispanic whites, in the Midwest and West, but not other regions. In the Northeast only, Blacks had 2-fold higher multivariable-adjusted odds, compared to non-Hispanic whites, of both core depression features (anhedonia and sadness) and appetite problems. There was also regional variation in neurovegetative features. For example, compared to non-Hispanic whites, Hispanics had 3–4-fold higher multivariable-adjusted odds of appetite disturbance in the Midwest and 4–5-fold higher odds of psychomotor symptoms in the Southeast, but there were no such differences in other regions. FDR correction was performed by item-level symptoms; 8 different FDR corrections were made. In each item-level specific FDR correction, 16 tests were included (i.e., 4 racial/ethnic subgroup comparisons in all 4 regions).

Geographic region and depression care, including racial/ethnic differences by region

Table 4A addresses the likelihood of receiving depression care among those who reported clinically significant depressive symptoms (PHQ-8≥10); due to lower numbers of participants in this analytic subset, we combined Hispanic, Asian, and Other race/ethnicity participants into one category. There was no significant difference in the likelihood of receiving depression care by geographic region, although estimates were in the direction of ~20–25% lower odds of receiving treatment in all regions when compared to the Northeast. Previously, when addressing the entire sample regardless of region, we observed 60%-80% lower odds of depression care among Blacks compared to non-Hispanic whites.3 When stratifying by region, we found comparable estimates of 60–80% lower odds of depression care among Black compared to non-Hispanic whites in all regions – except in the Northeast, where the point estimate was slightly lower but not statistically significant.3 No regional differences in depression care were observed among racial/ethnic minorities compared to non-Hispanic whites.

Table 4A.

Adjusted Odds Ratio (OR) for Depression Care by US Geographic Region and Patterns of Racial/ethnic Differences in Depression Care within a Geographic Region, Among Those Reporting Clinically Significant Depressive Symptoms (PHQ-8 ≥ 10)

Outcome (n=755) US Geographic region Comparison: Region vs. LLD carea Comparison: Race/ethnicity vs. LLD care within each geographic regionb
Non-Hispanic white Black Hispanic, Asian, and Other race/ethnicity
Adjusted OR (95% CI)d
Medication/counseling usec Northeast 1.00 (Ref) 1.00 (Ref) 0.53 (0.26 – 1.06) 2.16 (0.76 – 6.11)
Southeast 0.80 (0.54 – 1.19) 1.00 (Ref) 0.37 (0.20 – 0.70) 1.03 (0.44 – 2.38)
Midwest 0.74 (0.46 – 1.17) 1.00 (Ref) 0.21 (0.07 – 0.60) 0.84 (0.22 – 3.14)
West 0.72 (0.46 – 1.13) 1.00 (Ref) 0.27 (0.09 – 0.80) 1.02 (0.40 – 2.58)
Open in a new tab

In Table 4B, we separately assessed depression care among participants who reported clinician-diagnosed depression; we observed ~15–20% lower odds of depression care in the West and Midwest, compared to the Northeast, although estimates were statistically significant only in the Midwest. Table 4B also shows that, even among those with clinician-diagnosed depression, Blacks had a statistically significant ~40% lower odds of receiving any depression care than non-Hispanic whites across all regions. Of note, we included a total of 8 tests for FDR correction (i.e., 2 racial/ethnic subgroup comparisons in all 4 regions); separate testing was performed for Tables 4A and 4B.

Table 4B.

Adjusted Odds Ratio (OR) for Depression Care by US Geographic Region and Patterns of Racial/ethnic Differences in Depression Care within a Geographic Region, Among Those Reported Clinician/physician Diagnosed Depression in Past 2 Years

Outcome (n=4846) US Geographic region Comparison: Region vs. LLD carea Comparison: Race/ethnicity vs. LLD care within each geographic regionb
Non-Hispanic white Black Hispanic, Asian, and Other race/ethnicity
Adjusted OR (95% CI)d
Medication/counseling usec Northeast 1.00 (Ref) 1.00 (Ref) 0.60 (0.43 – 0.84) 1.69 (1.08 – 2.63)
Southeast 0.95 (0.81 – 1.11) 1.00 (Ref) 0.60 (0.44 – 0.81) 1.10 (0.73 – 1.66)
Midwest 0.84 (0.71 – 1.00) 1.00 (Ref) 0.59 (0.41 – 0.85) 1.12 (0.68 – 1.86)
West 0.76 (0.65 – 0.90)* 1.00 (Ref) 0.62 (0.41 – 0.95) 1.37 (0.96 – 1.97)
Open in a new tab

Abbreviation: PHQ, patient health questionnaire; OR, odds ratio; CI, confidence interval

Variable definitions are provided in the footnotes of Table 1.

a

The northeast region was used as a reference group for the association of geographic region and depression care.

*

Wald chi-square test, df=1, p-value <0.05.

b

Non-Hispanic white participants were reference group for patterns of racial/ethnic differences in the depression care within a geographic region. Separate FDR corrections were performed for Table 4A and Table 4B. Estimates that remained significant after multiple hypothesis testing using a Wald chi-square test (df=1) at false discovery rate (FDR)<0.05 are highlighted in bold. A total of 8 FDR correction tests were included (i.e., 2 racial/ethnic subgroup comparisons in all 4 regions); separate testing was performed for Tables 4A and 4B.

c

Medication/counseling use was determined based on self-reported use of selective serotonin reuptake inhibitors (SSRIs) or other medications for depression and/or counseling for depression.

d

The ORs shown were adjusted for demographic, lifestyle/behavioral, and comorbidity factors. For both analyses, to avoid quasi-separation issues/undefined estimates in the adjusted models, we imputed the mean value to missing body mass index, the median value to physical activity, and the largest education category to that missing education information. We combined some categories to create binary variables: smoking use (ever/never smoker) and alcohol frequency use (daily vs. non-daily use). For comorbidity variables such as history of hypertension, diabetes, and high cholesterol: missing participants were combined to reference (‘no’) categories.

DISCUSSION

LLD severity varied significantly by geographic region; severity levels were lower in the Midwest than in other regions. Despite a lower overall depression severity and item-level symptom burden in the Midwest, racial/ethnic disparities in LLD were most pronounced in this region after adjusting for numerous covariates. For example, Hispanic participants had over 3-fold higher adjusted odds of anhedonia, sadness, and appetite problems compared to non-Hispanic whites in the Midwest, but not in other regions; in the Northeast, Blacks compared to non-Hispanic whites had 2-fold higher odds of both core depression features (anhedonia and sadness). Among those who self-reported elevated symptoms (PHQ-8≥10), no regional differences in depression care were observed, although estimates were in the direction of lower odds in all regions compared to the Northeast. Across all regions, Blacks were observed to have 50% or higher likelihood of not receiving any depression care, than non-Hispanic whites.

Although differences in LLD severity by geographic region have been reported previously,4 to our knowledge, this is the first study to show differences in item-level symptom burden by region. Consistent with our findings, a CDC (Centers for Disease Control and Prevention) report found that, of the top ten US states with the best mental health (i.e., lower percentages of poor mental health days in the past 30 days), seven were in the Midwest.21 Lower LLD care in all regions vs. the Northeast may be explained by the relative shortages of general and/or geriatric psychiatrists and inadequate access to health care in regions other than the Northeast.22

Patterns of racial/ethnic disparities in LLD were not uniform across US regions. In the Northeast, Black and Other/Multiple/Unspecified-race participants had higher LLD severity levels compared to non-Hispanic whites; in the Midwest, Black and Hispanic participants had higher LLD severity compared to non-Hispanic whites. However, no racial/ethnic disparities in LLD severity were observed in the Southeast and West. Consistent with these findings, data from the CDC suggested that Hispanics in the Midwest and Blacks in the Northeast reported a higher average number of poor mental health days in the past 30 days than non-Hispanic whites in those respective regions.23 It has been reported that Northeasterners and Midwesterners may have higher overall levels of stress than Southerners and Westerners;24 other data have indicated that minority adults, especially Black and Hispanic, have a higher burden of chronic stress and stress-sensitive adverse health behaviors that increase the risk of depression.25,26 Thus, a speculative possibility is that higher regional stress levels among minorities could explain the observed racial/ethnic disparities in LLD by region.

These results may also be considered in the context of a reported Black-White mental health “paradox”27,28 – i.e., the apparent “paradox” of lower rates of depression diagnoses among Blacks compared to non-Hispanic whites, despite higher prevalence of key risk factors (e.g., medical co-morbidities, life stressors). However, while prior studies suggesting this paradox included all adults aged ≥18 years, our study focused on older adults and results may not be directly comparable. Indeed, our results showing higher depressive symptom severity among older Black and minority adults are consistent with a prior study.29 Other investigators have highlighted a potential “double paradox,”30 referring to the evidence that Black adults report higher levels of psychological distress (e.g., on symptom scales) and yet have lower rates of criteria-based depression diagnoses – suggesting possible measurement or classification issues warranting further investigation. Our study suggests that geographic region should be considered in future work addressing these apparent paradoxes.

This study makes a novel contribution to the literature by reporting geographic variation in item-level symptom burden and racial/ethnic disparities therein. Racial/ethnic disparities in symptom burden were observed in several regions; however, the magnitude of disparity, especially in Black, Hispanic and Asian participants compared to non-Hispanic whites, was most pronounced in the Midwest. Further, lower LLD care among Blacks compared to non-Hispanic whites is apparent across all regions, although the magnitude of this disparity appeared less pronounced in the Northeast. In an analysis involving n=3,211 participants, Kim and co-workers had reported that in the Northeast region there were no significant racial/ethnic differences in unmet mental health needs (defined as either not using mental health services or using services inadequately while having symptoms of mental disorders), whereas care disparities were apparent in other regions.8 Our study extends these findings by addressing depression care disparities among over 25,000 older adults.

Our results indicate a critical need for further examination of regional variation in social and health determinants that may have influenced our findings of racial/ethnic disparities in LLD by region. Although we adjusted for several known social and health determinants (e.g., household income, education, physical activity, smoking, alcohol use, comorbid conditions), the observed disparities in LLD outcomes might have been influenced by unmeasured social determinants of health (e.g., health insurance coverage;31 social isolation;32,33 neighborhood deprivation;34 racial bias or discrimination;31,35 access to care;36 food insecurity;37 patient-physician/clinician communication problems38) that may vary in prevalence by geographic region. Moreover, the estimated racial/ethnic differences in odds of depression care across regions differed, depending on whether the analysis was conditioned on self-reported clinician-diagnosed depression vs. participants’ own self-reported symptoms (PHQ-8≥10). In the latter analysis, there were stronger point estimates indicating lower odds of depression care among Blacks than non-Hispanic whites; overlap in the confidence bounds in the two sets of analyses should be noted. Nevertheless, a potential explanation for the differences in estimates is lower sensitivity of physicians/clinicians across all geographic regions in recognizing the clinical importance of depressive symptoms, with subsequent under-treatment of depression, among Black compared to non-Hispanic white patients. Future studies are needed to explore the unmeasured patient and provider factors involved in depression care and decision-making.

Notably, we found a lower overall depression severity and symptom burden in the Midwest compared to other regions – yet significantly higher LLD severity and symptom burden among minorities compared to non-Hispanic whites within the Midwest. We considered potential explanations. First, although the models were adjusted for numerous LLD risk factors, it is possible that regional differences in unmeasured confounders, including social and health determinants (noted above), may partly explain findings. Second, over the past decade, Hispanic persons have accounted for ~60% of the population growth in the Midwest, including among recent immigrants, displaced persons, and those seeking refuge;39 future research might address this potential source of heterogeneity in observed larger LLD disparities among Hispanic adults in the Midwest.

Our study has clinical and research implications. First, these findings may inform design of future studies, by highlighting the importance of measuring geographic region and social determinants of health linked to region when investigating racial/ethnic disparities in LLD. Second, the novel finding of greater magnitude of racial disparity in LLD outcomes in some regions compared to others highlights the need for rigorous investigations of mechanistic drivers. For example, regional differences in neighborhood-level factors or in manifestations of structural and/or interpersonal racism, bias or discrimination may lead to higher expressions of dysphoria, low self-esteem, or other item-specific symptoms – as well as overall high depression severity – among minorities in those regions. Knowledge of such regional variation could inform region-specific LLD prevention strategies.

This study had several strengths. First, the sample is large, racially and ethnically diverse, and well-characterized with respect to demographic, lifestyle/behavioral, and health factors. Second, the study makes a novel contribution to existing literature, as it not only addresses differences in LLD outcomes by region but also helped to de-convolute the role of geographic region in observed patterns of racial/ethnic disparities in LLD. Third, to our knowledge, prior reports do not address associations between geographic region and item-level symptom burden or examined racial/ethnic disparities in symptom burden by region.

We acknowledge limitations. First, the study is cross-sectional; a longitudinal approach would provide clarity regarding potential racial/ethnic differences in prospective change in LLD outcomes by region. Thus, a temporal relationship between exposure and outcome cannot be determined. Second, potential generalizability concerns are noted. Most VITAL participants were healthy at baseline. Also, although the overall racial/ethnic distribution in VITAL at baseline (70% non-Hispanic white, 30% minority) was similar to US Census data at that time (72% non-Hispanic white, 28% minority),40 regional distributions of racial/ethnic groups may have differed (e.g., higher representation of Blacks in all regions and lower representation of Hispanics in the West in VITAL vs. US Census data). However, these generalizability issues would not affect internal validity of findings. Third, although we adjusted for numerous confounders, we did not collect information on other social determinants and psychosocial factors. Fourth, the higher point estimates for item-level depressive symptom burden among Hispanics and Asians in the Midwest were notable but should be interpreted with caution, considering the relatively smaller numbers of Hispanic and Asian participants in the sample. Fifth, misclassification of race/ethnicity is possible; however, random misclassification if present would tend to bias estimates towards rather than away from the null and render findings more conservative.

In summary, we observed significant regional differences in LLD severity and item-level symptom burden, but not in treatment. Lower levels of LLD severity and item-level symptom burden were observed in the Midwest compared to the Northeast. Furthermore, geographic region may play an important role in patterns of racial/ethnic disparities in LLD outcomes. Future depression prevention studies may need to account for geographic region as an explanatory factor when addressing racial/ethnic disparities in LLD.

Supplementary Material

Supplement

Acknowledgements:

We are indebted to the 25,871 VITAL participants and the entire VITAL staff for their dedicated and conscientious collaboration, with special appreciation for Ms. Alison Weinberg and Ms. Jennifer Coates for assistance with the VITAL-DEP ancillary study.

Funding:

This work was supported by R01 MH091448 from National Institute of Mental Health (NIMH). VITAL is supported by grants U01 CA138962, R01 CA138962 and R01 AT011729, which includes support from the National Cancer Institute; National Heart, Lung and Blood Institute (NHLBI); Office of Dietary Supplements; National Institute of Neurological Disorders and Stroke; and the National Center for Complementary and Integrative Health of the National Institutes of Health (NIH). The VITAL ancillary studies and CTSC component are supported by grants DK088078 and R01 DK088762 from the National Institute of Diabetes and Digestive and Kidney Diseases; R01 HL101932 and R01 HL102122 from NHLBI; R01 AG036755 from the National Institute on Aging (NIA); R01 AR059086 and R01 AR060574 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases; and R01 MH091448 from the National Institute of Mental Health (NIMH). Dr. Reynold’s participation is also supported by P30 MH090333 from NIMH and the University of Pittsburgh Medical Center Endowment in Geriatric Psychiatry. Pharmavite LLC of Northridge, California (vitamin D) and Pronova BioPharma (BASF) of Norway (Omacor® fish oil) donated the study agents, matching placebos, and packaging in the form of calendar packs.

Sponsor’s role:

The funding sources had no role in the study design, data analysis or interpretation, or the preparation of the manuscript.

DISCLOSURES

Dr. Mischoulon has received research support from Nordic Naturals and Heckel medizintechnik GmbH. He has received honoraria for speaking from the Massachusetts General Hospital Psychiatry Academy, Harvard Blog, and PeerPoint Medical Education Institute, LLC. He also works with the MGH Clinical Trials Network and Institute (CTNI), which has received research funding from multiple pharmaceutical companies and NIMH.

Dr. Okereke receives royalties from Springer Publishing for a book on late-life depression prevention.

Dr. Reynolds receives payment from the American Association of Geriatric Psychiatry as Editor-in-Chief of the American Journal of Geriatric Psychiatry, royalty income for intellectual property as co-inventor of the Pittsburgh Sleep Quality Index, and honorarium from Merck for consultation on care pathways for insomnia.

Dr. Chang receives royalties from Up-to-Date.

Footnotes

Previous presentation: The Gerontological Society of America 2020 Annual Scientific Meeting, Virtual Conference, November 4–7, 2020.

REFERENCES

  • 1.Alegría M, Chatterji P, Wells K, et al. Disparity in depression treatment among racial and ethnic minority populations in the United States. Psychiatr Serv. 2008;59(11):1264–1272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Akincigil A, Olfson M, Siegel M, et al. Racial and ethnic disparities in depression care in community-dwelling elderly in the United States. Am J Public Health. 2012;102(2):319–328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Vyas CM, Donneyong M, Mischoulon D, et al. Association of race and ethnicity with late-life depression severity, symptom burden, and care. JAMA Netw Open. 2020;3(3):e201606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Killian TS, Penfield M. Predictors of depressive symptoms: What are the roles of geography and informal social support? Adv Appl Sociol 2012;2:313–319. [Google Scholar]
  • 5.Henriquez-Sanchez P, Doreste-Alonso J, Martinez-Gonzalez MA, et al. Geographical and climatic factors and depression risk in the SUN project. Eur J Public Health. 2014;24(4):626–631 [DOI] [PubMed] [Google Scholar]
  • 6.Kim G, Parton JM, DeCoster J, et al. Regional variation of racial disparities in mental health service use among older adults. Gerontologist. 2013;53(4):618–626. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Kim G, Parton JM, Ford KL, et al. Geographic and racial-ethnic differences in satisfaction with and perceived benefits of mental health services. Psychiatr Serv. 2014;65(12):1474–1482. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Kim G, Dautovich N, Ford KL, et al. Geographic variation in mental health care disparities among racially/ethnically diverse adults with psychiatric disorders. Soc Psychiatry Psychiatr Epidemiol. 2017;52(8):939–948 [DOI] [PubMed] [Google Scholar]
  • 9.World Health Organization, Media Centre, Depression Fact Sheet, Updated February 2017.
  • 10.Strine TW, Mokdad AH, Balluz LS, et al. Depression and anxiety in the United States: findings from the 2006 Behavioral Risk Factor Surveillance System. Psychiatr Serv. 2008;59(12):1383–1390 [DOI] [PubMed] [Google Scholar]
  • 11.Trust for America's Health. Examples of successful community-based public health interventions (state-by-state). https://www.tfah.org/wp-content/uploads/2018/09/Examplesbystate1009.pdf. Accessed May 17, 2020.
  • 12.Chang SC, Wang W, Pan A, et al. Racial variation in depression risk factors and symptom trajectories among older women. Am J Geriatr Psychiatry. 2016;24(11):1051–1062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Howard VJ, Cushman M, Pulley L, et al. The reasons for geographic and racial differences in stroke study: objectives and design. Neuroepidemiology. 2005;25(3):135–143 [DOI] [PubMed] [Google Scholar]
  • 14.Howard VJ, McClure LA, Meschia JF, et al. High prevalence of stroke symptoms among persons without a diagnosis of stroke or transient ischemic attack in a general population: the REasons for Geographic And Racial Differences in Stroke (REGARDS) study. Arch Intern Med. 2006;166(18):1952–1958 [DOI] [PubMed] [Google Scholar]
  • 15.Manson JE, Cook NR, Lee IM, et al. Vitamin D supplements and prevention of cancer and cardiovascular disease. N Engl J Med. 2019;380(1):33–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Okereke OI, Reynolds CF 3rd, Mischoulon D, et al. Effect of long-term vitamin D3 supplementation vs placebo on risk of depression or clinically relevant depressive symptoms and on change in mood scores: A Randomized Clinical Trial. JAMA. 2020;324(5):471–480. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Kroenke K, Strine TW, Spitzer RL, et al. The PHQ-8 as a measure of current depression in the general population. J Affect Disord. 2009;114(1–3):163–173 [DOI] [PubMed] [Google Scholar]
  • 18.McGuire LC, Strine TW, Allen RS, et al. The Patient Health Questionnaire 8: current depressive symptoms among U.S. older adults, 2006 Behavioral Risk Factor Surveillance System. Am J Geriatr Psychiatry. 2009;17(4):324–334 [DOI] [PubMed] [Google Scholar]
  • 19.Diez-Quevedo C, Rangil T, Sanchez-Planell L, et al. Validation and utility of the patient health questionnaire in diagnosing mental disorders in 1003 general hospital Spanish inpatients. Psychosom Med. 2001;63(4):679–686 [DOI] [PubMed] [Google Scholar]
  • 20.Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the royal statistical society Series B (Methodological). 1995:289–300. [Google Scholar]
  • 21.Centers for Disease Control and Prevention (CDC). Behavioral Risk Factor Surveillance System Survey Data. Atlanta, Georgia: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, [2016]. [Google Scholar]
  • 22.Beck AJ, Page C, Buche MJ, Rittman D, Gaiser M. Estimating the distribution of the US psychiatric subspecialist workforce. Population. 2018. Dec 2;600:47–6. [Google Scholar]
  • 23.Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Division of Population Health. Chronic Disease Indicators (CDI) Data [online]. 2016. [accessed May 13, 2020]. URL: https://nccd.cdc.gov/cdi.
  • 24.Stress and Religion [press release]. American Psychological Association. 2011. URL: https://www.apa.org/news/press/releases/stress/2011/region.
  • 25.Jackson JS, Knight KM, Rafferty JA. Race and unhealthy behaviors: chronic stress, the HPA axis, and physical and mental health disparities over the life course. Am J Public Health. 2010;100(5):933–939. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Rodriquez EJ, Gregorich SE, Livaudais-Toman J, et al. Coping with chronic stress by unhealthy behaviors: A re-evaluation among older adults by race/ethnicity. J Aging Health. 2017;29(5):805–825. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Erving CL, Thomas CS, Frazier C. Is the Black-White mental health paradox consistent across gender and psychiatric disorders? Am J Epidemiol. 2019;188(2):314–322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Gibbs TA, Okuda M, Oquendo MA, et al. Mental health of African Americans and Caribbean blacks in the United States: results from the National Epidemiological Survey on Alcohol and Related Conditions. Am J Public Health. 2013;103(2):330–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Barry LC, Thorpe RJ Jr., Penninx BW, et al. Race-related differences in depression onset and recovery in older persons over time: the health, aging, and body composition study. Am J Geriatr Psychiatry. 2014;22(7):682–691. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Barnes DM, Bates LM. Do racial patterns in psychological distress shed light on the Black-White depression paradox? A systematic review. Soc Psychiatry Psychiatr Epidemiol. 2017;52(8):913–928 [DOI] [PubMed] [Google Scholar]
  • 31.McGuire TG, Miranda J. New evidence regarding racial and ethnic disparities in mental health: policy implications. Health Aff (Millwood). 2008;27(2):393–403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Miyawaki CE. Association of social isolation and health across different racial and ethnic groups of older Americans. Ageing Soc. 2015;35(10):2201–2228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Taylor HO, Taylor RJ, Nguyen AW, et al. Social isolation, depression, and psychological distress among older adults. J Aging Health. 2018;30(2):229–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Galea S, Ahern J, Nandi A, et al. Urban neighborhood poverty and the incidence of depression in a population-based cohort study. Ann Epidemiol. 2007;17(3):171–179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Balsa AI, McGuire TG. Prejudice, clinical uncertainty and stereotyping as sources of health disparities. J Health Econ. 2003;22(1):89–116 [DOI] [PubMed] [Google Scholar]
  • 36.Cummings JR, Wen H, Ko M, et al. Geography and the Medicaid mental health care infrastructure: implications for health care reform. JAMA Psychiatry. 2013;70(10):1084–1090. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Leung CW, Epel ES, Willett WC, et al. Household food insecurity is positively associated with depression among low-income supplemental nutrition assistance program participants and income-eligible nonparticipants. J Nutr. 2015;145(3):622–627 [DOI] [PubMed] [Google Scholar]
  • 38.Conner KO, Copeland VC, Grote NK, et al. Mental health treatment seeking among older adults with depression: the impact of stigma and race. Am J Geriatr Psychiatry. 2010;18(6):531–543. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Immigration in the Global Midwest: Immigration History Research Center / Unniversity of Minnesota; 2016. [Available from: https://cla.umn.edu/ihrc/research/immigration-global-midwest.
  • 40.Humes KR, Jones NA, Ramirez RR. Overview of Race and Hispanic origin: 2010. Washington, DC: 2011. Report No.: 2010 Census Briefs, C2010BR-02. URL: https://www.census.gov/prod/cen2010/briefs/c2010br-02.pdf [Google Scholar]

Associated Data

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

Supplementary Materials

Supplement
NIHMS1758337-supplement-Supplement.docx (409.4KB, docx)

RESOURCES