Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Nov 15.
Published in final edited form as: J Affect Disord. 2015 Aug 28;187:147–150. doi: 10.1016/j.jad.2015.08.002

Gene-by-social-environment interaction (GxSE) between ADCYAP1R1 genotype and neighborhood crime predicts major depression symptoms in trauma-exposed women

Sarah R Lowe a, John Pothen b, James W Quinn a, Andrew Rundle a, Bekh Bradley c,d, Sandro Galea e, Kerry J Ressler d, Karestan C Koenen a
PMCID: PMC4587299  NIHMSID: NIHMS718696  PMID: 26334183

Abstract

Background

Few studies have explored interactions between genes and social environmental exposures (GxSEs) for trauma-related psychopathology, including symptoms of posttraumatic stress (PTS) and major depression (MD). The extant literature suggests the possibility of a GxSE between the rs2267735 variant of the ADCYAP1R1 gene and neighborhood crime. The current study aimed to explore this possibility among a predominantly African American sample of trauma-exposed women.

Methods

Female participants (N = 1361) were recruited from a public hospital, and completed measures of PTS and MD symptoms and provided DNA samples. Participants’ home addresses were mapped onto 300 neighborhoods (2010 census tracts), and data on crime within neighborhoods was collected.

Results

Multilevel models detected a significant GxSE between rs2267735 and neighborhood crime for MD symptoms (p = .01). Having two copies of the risk (C) allele was associated with higher MD symptoms for participants living in high-crime neighborhoods.

Limitations

At least six limitations are noteworthy: (1) low statistical power; (2) use of self-report symptom inventories; (3) lack of information on symptom onset; (4) homogeneous sample from a single metropolitan area; (5) non-specific index of crime; and (6) use of census tracts to define neighborhoods.

Conclusions

The results provide further evidence of GxSEs for psychiatric outcomes among trauma-exposed populations. Further investigations of genetic factors for trauma-related psychopathology should include careful assessments of the social environment.

Keywords: Posttraumatic stress, major depression, genetic risk, social environment, gene-by-social-environment interactions, neighborhood crime, multilevel modeling

Introduction

Exposure to traumatic events is associated with a range of adverse psychiatric consequences, including symptoms of posttraumatic stress (PTS) and major depression (MD) (Lowe et al., 2015). Several risk factors for higher PTS and MD symptoms have been identified, including genetic variants (G; Dunn et al., 2015; Skelton et al., 2012), individual environmental exposures (E; e.g., child abuse; Carr et al., 2013), and social environmental exposures (SE; e.g., neighborhood segregation; Lee, 2009). A growing body of literature has investigated interactions between genes and individual environmental exposures (GxEs) as predictors of both outcomes (e.g., Dunn et al., 2015; Skelton et al., 2012). Comparatively fewer studies have explored the joint influence of genes and social environmental exposures (GxSEs; e.g., Uddin et al., 2010, 2011).

One gene that might be involved in GxSEs for PTS and MD symptoms is ADCYAP1R1. ADCYAP1R1 encodes PAC1, the receptor for pituitary adenylate cyclase-activating polypeptide (PACAP), which has been implicated in stress response processes (Hashimoto et al., 2006) and PTS symptoms in females (Ressler et al., 2011). Two prior studies found that, controlling for comorbid symptoms, the risk for higher PTS associated with the rs2267735 variant of ADCYAP1R1 was enhanced among women with more extensive trauma histories, whereas this GxE was non-significant for MD symptoms (Almli et al., 2013; Uddin et al., 2013). Another study suggested that GxSEs with ADCYAP1R1 might have broader effects on trauma-related psychopathology (Roman et al., 2014). In this study, signaling of PACAP in a stress-related limbic structure through PAC1 receptors was associated with increased anxious behaviors and weight loss, and decreased food intake among rodents in a chronic stress, versus control, condition.

Living in a high-crime neighborhood represents a chronic stressor for human populations, and has been found to increase risk for mood and anxiety disorders (Stockdale et al., 2007). A previous study documented a significant GxSE with neighborhood crime among a predominantly White sample of natural disaster survivors, such that the risk for PTSD associated with the serotonin protein gene promoter variant was enhanced for participants living in high-crime neighborhoods (Koenen et al., 2009). Yet, it remains unclear whether this relationship extends to ADCYAP1R1, survivors of a broader range of traumatic events, and ethnic minority samples.

In summary, the extant literature suggests the possibility of a GxSE between the rs2267735 variant of ADCYAP1R1 and neighborhood crime for PTS and MD symptoms. In the current study, we tested this possibility among a predominantly African American sample of trauma-exposed women.

Methods

Participants and Procedures

Participants were recruited from the primary care and obstetrics gynecology clinics at a public hospital in Atlanta, Georgia. Participants completed a survey that included measures of PTS and MD symptoms, administered verbally due to the relatively high rates of impaired literacy in the community served at the hospital. The battery took 30–90 minutes to complete, and participants were offered $15 for their participation. The appropriate Institutional Review Boards approved the study, and participants provided written informed consent.

The current study included 1361 trauma-exposed female participants who provided complete data on all study variables. The sample was majority African American (94.1%). On average, participants were 37.47-years old (SD = 13.30; Range: 18–74).

Measures

Lifetime Trauma Exposure

Participants completed a screener of lifetime trauma exposure, the Traumatic Events Inventory (TEI; Schwartz et al., 2005). Participants indicated whether they had ever experienced 14 potentially traumatic events (e.g., child physical abuse, serious accident or injury), and feelings of terror, horror, or helplessness during such events.

PTS symptoms

PTS symptoms were assessed via the Posttraumatic Symptom Scale-Interview Version (PSS-I; Foa et al., 1993). The PSS consists of 17 items corresponding PTSD symptoms as specified in the Diagnostic and Statistical Manual of Mental Disorders (4th edition, text revision; DSM-IV-TR (American Psychiatric Association, 2000). Participants rated the extent to which they experienced each symptom (e.g., “Have you been having recurrent bad dreams or nightmares about the trauma?”) over the prior two weeks from 0 (not at all) to 3 (5 or more times per week), and a symptom severity score was computed as the sum of all ratings. The PSS-I has been shown to have high internal consistency, inter-rater reliability, and validity (Foa and Tolin, 2000) (M = 13.12, SD = 12.40, Range: 0–51; α = .92).

MD symptoms

The Beck Depression Inventory-II (BDI; Beck et al., 1996) assessed MD symptoms. The BDI consists of 21 items corresponding to DSM-IV-TR symptoms of major depressive disorder. For each item, participants indicate which of four statements best describes the way they have been feeling over the past two weeks. Each statement has a corresponding score, ranging from 0 (e.g., “I do not feel sad”) to 3 (e.g., “I am so sad or unhappy that I can’t stand it”). A symptom severity score was computed as the sum of all ratings. The BDI-II has had evidence of reliability, internal consistency, and validity across a wide variety of samples, including primary care medical patients (Arnau et al., 2001) (M = 14.91, SD = 12.36, Range: 0–58; α = .93).

Neighborhood crime

Participants’ geocoded home addresses were mapped onto 2010 census tracts (n = 300) and data on crime within tracts were collected from the CrimeRisk 2010 (Applied Geographic Solutions, 2010). Values are relative to the national average, which is set at 100. For example, a value of 200 would mean that the tract had twice the amount of crime as the national average, and a value of 50 would mean that the tract had half the amount of crime as the national average. Crimes included within this index are murder, rape, assault, robbery, burglary, larceny, and motor vehicle theft. The crime variable was divided by its interquartile range at the community-level (IQR = 344) prior to entry into the analysis to facilitate interpretation. In the analysis, each unit increase in crime is therefore equivalent to the difference between the 25th percentile neighborhood and the 75th percentile neighborhood (M = 267.53, SD = 234.54, Range: 5–892).

Genotyping

Participants provided DNA samples from blood or saliva samples. Further details on DNA collection have been published previously (masked for blind review). Briefly, samples were genotyped using both Sequenom (Sequenom Inc., San Diego, CA) and Taqman (Applied Biosystems Inc., Foster City, CA). The rs2267735 SNP passed quality control filters (call rate > 95%, minor allele frequency > 0.01, Hardy-Weinberg disequilibrium P > 1×10−6). Genotypic frequencies were as follows: G/G: 11.7% (n = 159); G/C: 46.1% (n = 627); C/C: 42.2% (n = 575). To be consistent with the two prior studies documenting GxEs with rs2267735 (Almli et al., 2013; Uddin et al., 2013), we assumed an additive genetic effect, such that the number of risk (C) alleles was included in statistical models.

Data Analysis

Analyses were conducted in Mplus 7.1 (Muthén & Muthén, 1998–2012). Prior to analyses fulfilling study aims, we estimated the bivariate association between rs2267735 and crime, which could potentially undermine the interpretation of a significant GxSE (Dick, 2011). The association was non-significant (Coeff. = .01, SE = .03, p = .69).

Subsequently, multilevel models predicting PTS and MD symptoms, with participants nested in census tracts, were conducted. Model 1 for each outcome included the main effects of the number of rs2267735 C alleles, demographics (age, self-reported African American ethnicity) and comorbid symptoms at the individual-level, and crime at the neighborhood-level. Comorbid symptoms were MD symptoms in the model predicting PTS symptoms, and PTS symptoms in the model predicting MD symptoms. Model 2 for each outcome included the addition of a cross-level interaction between the number of risk alleles and crime. We present nominally significant results (p < .05) and note whether they held with a Bonferroni correction for the number of outcomes being analyzed (α/2; p < .025). Unstandardized results are listed, as standardized coefficients are unavailable in multilevel modeling in Mplus 7.1.

Results

Table 1 shows the results of the multilevel analyses. For PTS, having more copies of the rs2267735 C allele was a nominally significant predictor of higher PTS (p = .04), and higher MD symptoms a significant predictor of higher PTS symptoms after Bonferroni correction (p < .001) in Model 1. In Model 2, the cross-level association between rs2267735 and neighborhood crime was non-significant (p = .44).

Table 1.

Results of Multilevel Level Models Predicting Posttraumatic Stress and Major Depression Symptoms

PTS Symptoms MD Symptoms
Model 1 Model 2 Model 1 Model 2
Coeff. (SE) Coeff. (SE) Coeff. (SE) Coeff. (SE)
Individual-level
  Age −02 (.02) −02 (.02) .06 (.02)*** .05 (.03)
  African American −1.78 (1.17) −1.84 (1.17) −46 (1.08) −1.37 (2.73)
  Comorbid symptoms .71 (.02)*** .71 (.02))*** .70 (.02))*** .69 (.03))***
  rs2267735 .74 (.36)* 1.15 (.63) −38 (.32) −1.46 (1.45)
Community-level
  Crime .10 (.38) .10 (.38) −06 (.35) −09 (.73)
Cross-level
  rs2267735×crime -- −42 (.54) -- 1.20 (.48)*
Open in a new tab

Note. N = 1361 female participants nested in 300 neighborhoods (2010 census tracts). PTS = Posttraumatic Stress, MD = Major Depression, Coeff. = Coefficient, SE = Standard Error.

Results in boldface held after a Bonferroni-correction of α/2 (p < .025).

*

p < .05,

**

p < .01,

***

p < .001

For MD symptoms, older age and higher PTS were significant predictors of higher MD symptoms after Bonferroni correction (both p < .001) in Model 1. In Model 2, the cross-level interaction between rs2267735 and crime was statistically significant after Bonferroni correction (p = .01). To illustrate the interaction, mean levels of MD symptoms were computed in the individual-level data file for participants 0, 1, and 2 copies of the rs2267735 C allele, living in neighborhoods with high- versus low-levels of crime (defined as one standard deviation or more above or below the mean, respectively). As shown in Fig. 1, having two copies of the C allele was associated with risk for higher MD symptoms for participants living in high-crime neighborhoods.

Figure 1. Decomposition of cross-level interaction between neighborhood crime and rs2267735 in predicting major depression symptoms.

Figure 1

Open in a new tab

N = 1361 female participants living in 300 neighborhoods (2010 census tracts). MD = major depression. High and low crime neighborhoods are defined as one standard deviation or more above or below the mean, respectively.

Discussion

We found a significant GxSE between the rs2267735 variant of the ADCYAP1R1 gene and neighborhood crime for MD symptoms among trauma-exposed women. Higher MD symptoms were observed among women carrying two copies of the C allele who were living in high-crime neighborhoods.

The results aligned with prior research showing GxEs including this variant for trauma-related psychopathology (Almli et al., 2013; Uddin et al., 2013). They extend this research by using a social environmental exposure, showing a significant GxSE for MD symptoms. The pattern of results was inconsistent with the previous study showing a GxSE with neighbourhood crime to be predictive of PTS symptoms (Koenen et al., 2009). The discrepancy is likely due to differences between the two samples (e.g., in demographic characteristics, trauma histories), as well as in the genetic variants included.

Future research is needed to replicate the results and provide an understanding of the biological mechanisms that underlie them. For example, prior research has implicated reactivity of the hippocampus and amygdala, and functional connectivity between these structures, as mediating the relationship between ADCYAP1R1 and women’s trauma-related psychopathology (Stevens et al., 2014). Future investigations could explore these processes among women living in high- and low-crime neighborhoods. Additionally, further analyses could explore GxSEs with other social environmental exposures that have been shown to influence trauma-related psychopathology in African Americans, including neighborhood racial composition and poverty (e.g., Dallaire et al., 2008; Wright et al., 2005). Taken together, these lines of research could help identify African American women most likely to experience trauma-related psychopathology based on the combined presence of genetic and social environmental risks, and inform efforts to prevent and reduce trauma-related symptoms through biological and psychosocial interventions.

At least six limitations of the study are worth noting. First, the sample was relatively small, limiting statistical power. Second, symptoms were assessed using self-report inventories, which are not substitutable for clinician assessments. Third, we did not have information on symptom onset, for example whether onset occurred preceded or followed participants’ residence in their current neighborhoods. Fourth, the study was conducted in a homogeneous sample. However, this sample has high rates of trauma-related psychopathology, increasing the public health relevance of the research (Gillespie et al., 2009). Ethnic homogeneity was also advantageous in the current study, given that genetic variants have been associated with different patterns of risk among different ethnic groups and that population stratification can lead to spurious GxE findings (Nugent et al., 2011). Fourth, the index of crime included a range of personal and property crimes, and it is unknown whether one or more of these drove the significant cross-level interaction. Lastly, neighborhoods were defined by census tracts, which perhaps do not reflect participants’ perceived neighborhoods.

Despite these limitations, these findings contribute to research on genetic and environmental risk for trauma related psychopathology. To date very few genetic studies examine social environment, despite its importance in risk for trauma related psychopathology. The results provide further evidence that social environmental exposures can shape genetic risk for psychiatric outcomes among trauma-exposed populations. Further investigations of genetic factors for trauma-related psychopathology should therefore include careful assessments of the social environment.

Research Highlights.

  • Genes and the social environment shape risk for trauma-related psychopathology.

  • Few studies have explored gene-by-social-environment interactions (GxSEs).

  • Female participants (N = 1361) completed surveys and provided DNA samples.

  • A significant GxSE between the ADCYAP1R1 gene and neighborhood crime was found.

  • Genetic risk for depression was enhanced for women in high-crime neighborhoods.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Almli LM, Mercer KB, Kerley K, Feng H, Bradley B, Conneely KN, Ressler KJ. ADCYAP1R1 genotype associates with post-traumatic stress symptoms in highly traumatized African-American females. Am. J. Med. Genet. B Neuropsychiatr. Genet. 2013;162B:262–272. doi: 10.1002/ajmg.b.32145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. American Psychological Association. Diagnostic and statistical manual of mental disorders, fourth ed., text rev. Washington, DC: Author; 2000. [Google Scholar]
  3. Applied Geographic Solutions. [Accessed February 23, 2015];Database methodology guide, 2010A update [Online] < http://www.appliedgeographic.com/ags_2010%20web%20pdf%20files/ags2010ametho dologyguide.pdf>.
  4. Arnau RC, Meagher MW, Norris MP, Bramson R. Psychometric evaluation of the Beck Depression Inventory-II with primary care medical patients. Health Psychol. 2001;20:112–119. doi: 10.1037//0278-6133.20.2.112. [DOI] [PubMed] [Google Scholar]
  5. Beck AT, Steer RA, Brown G. Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation; 1996. [Google Scholar]
  6. Carr CP, Martins CMS, Stingel AM, Lemgruber VB, Juruena MF. The role of early life stress in adult psychiatric disorders: A systematic review according to childhood trauma subtypes. J. Nerv. Ment. Dis. 2013;201:1007–1020. doi: 10.1097/NMD.0000000000000049. [DOI] [PubMed] [Google Scholar]
  7. Dallaire DH, Cole DA, Smith TM, Ciesla JA, LaGrange B, Jacquez FM, Pineda AQ, Truss AE, Folmer AS. Predicting children’s depressive symptoms from community and individual factors. J. Youth Adolesc. 2008;37:830–846. doi: 10.1007/s10964-008-9270-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Dick DM. Gene-environment interaction in psychological traits and disorder. Annu. Rev. Clin. Psychol. 2011;7:383–409. doi: 10.1146/annurev-clinpsy-032210-104518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dunn EC, Brown RC, Dai Y, Rosand J, Nugent NR, Amstadter AB, Smoller JW. Genetic determinants of depression: Recent findings and future directions. Harv. Rev. Psychiatry. 2015;23:1–18. doi: 10.1097/HRP.0000000000000054. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Foa EB, Riggs DS, Dancu CV, Rothbaum BO. Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. J. Trauma. Stress. 1993;6:459–473. [Google Scholar]
  11. Foa EB, Tolin DF. A comparison of the PTSD Symptom Scale-Interview Version and the Clinician-Administered PTSD Scale. J. Trauma. Stress. 2000;13:181–191. doi: 10.1023/A:1007781909213. [DOI] [PubMed] [Google Scholar]
  12. Gillespie CF, Bradley B, Mercer K, Smith AK, Connely K, Gapen M, Weiss T, Schwartz AC, Cubells JF, Ressler KJ. Trauma exposure and stress-related disorders in inner city primary care patients. Gen. Hosp. Psychiatry. 2009;31:505–514. doi: 10.1016/j.genhosppsych.2009.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hashimoto H, Shintani N, Baba A. New insights into the central PACAPergic system from the phenotypes in PACAP- and PACAP receptor-knockout mice. Ann. N. Y. Acad. Sci. 2006;1070:75–89. doi: 10.1196/annals.1317.038. [DOI] [PubMed] [Google Scholar]
  14. Koenen KC, Aiello AE, Bakshis E, Amstadter AB, Ruggiero KJ, Acierno R, Kilpatrick DG, Gelernter J, Galea S. Modification of the association between serotonin transporter genotype and risk of posttraumatic stress disorder in adults by county-level social environment. Am. J. Epidemiol. 2009;169:704–711. doi: 10.1093/aje/kwn397. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Koenen KC, Uddin M, Amstadter AB, Galea S. Incorporating the social environment in genotype environment interaction studies of mental disorders. Int. J. Clin. Pract. 2010;64:1489–1492. doi: 10.1111/j.1742-1241.2010.02359.x. [DOI] [PubMed] [Google Scholar]
  16. Lee M-A. Neighborhood residential segregation and mental health: A multilevel analysis on Hispanic Americans in Chicago. Soc. Sci. Med. 2009;68:1975–1984. doi: 10.1016/j.socscimed.2009.02.040. [DOI] [PubMed] [Google Scholar]
  17. Lowe SR, Blachman-Forshay J, Koenen KC. Epidemiology of trauma and traumarelated disorders: Trauma as a public health issue. In: Schnyder U, Cloitre M, editors. Evidence-based treatments for trauma-related psychological disorders: A practical guide for clinicians. New York: Springer; 2015. pp. 11–40. [Google Scholar]
  18. Muthén LK, Muthén BO. Mplus user’s guide. seventh ed. Los Angeles: Muthén & Muthén; 1998–2012. [Google Scholar]
  19. Nugent NR, Tyrka AR, Carpenter LL, Price LH. Gene-environment interactions: early life stress and risk for depressive and anxiety disorders. Psychopharmacology. 2011;214:175–196. doi: 10.1007/s00213-010-2151-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Ressler KJ, Mercer KB, Bradley B, Jovanovic T, Mahan A, Kerley K, Norrholm SD, Kilaru V, Smith AK, Myers AJ, Ramirez M, Engel A, Hammack SE, Toufexis D, Braas KM, Binder EB, May V. Post-traumatic stress disorder is associated with PACAP and the PAC1 receptor. Nature. 2011;470:492–497. doi: 10.1038/nature09856. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Roman CW, Lezak KR, Hartsock MJ, Falls WA, Braas KM, Howard AB, Hammack SE, May V. PAC1 receptor antagonist in the bed nucleus of the stria terminalis (BNST) attenuates the endocrine and behavioral consequences of chronic stress. Psychoneuroendocrinology. 2014;47:151–165. doi: 10.1016/j.psyneuen.2014.05.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Schwartz AC, Bradley RG, Sexton M, Sherry A, Ressler KJ. Posttraumatic stress disorder among African Americans in an inner city mental health clinic. Psychiatr. Serv. 2005;56:212–215. doi: 10.1176/appi.ps.56.2.212. [DOI] [PubMed] [Google Scholar]
  23. Skelton K, Ressler KJ, Norrholm SD, Jovanovic T, Bradley-Davino B. PTSD and gene variants: New pathways and new thinking. Neuropharmacology. 2012;62:628–637. doi: 10.1016/j.neuropharm.2011.02.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Stevens JS, Almli LM, Fani N, Guttman DA, Bradley B, Norrholm SD, Reiser E, Ely TD, Dhanani R, Glover EM, Jovanovic T, Ressler KJ. PACAP receptor gene polymorphism impacts fear responses in the amygdala and hippocampus. Proc. Natl. Acad. Sci. USA. 2014;111:158–163. doi: 10.1073/pnas.1318954111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Stockdale SE, Wells KB, Tang L, Belin TR, Zhang L, Sherbourne CD. The importance of social context: Neighborhood stressors, stress-buffering mechanisms, and alcohol, drug, and mental health disorders. Soc. Sci. Med. 2007;65:1867–1881. doi: 10.1016/j.socscimed.2007.05.045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Uddin M, Chang S-C, Zhang C, Ressler K, Mercer KB, Galea S, Keyes KM, McLaughlin KA, Wildman DE, Aiello AE, Koenen KC. ADCYAP1R1 genotype, posttraumatic stress disorder, and depression among women exposed to childhood maltreatment. Depress. Anxiety. 2013;30:251–258. doi: 10.1002/da.22037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Uddin M, de los Santos R, Bakshis E, Cheng C, Aiello AE. Building conditions, 5-HTTLPR genotype, and depressive symptoms in adolescent males and females. J. Adolesc. Health. 2011;49:379–385. doi: 10.1016/j.jadohealth.2011.01.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Uddin M, Koenen KC, de los Santos R, Bakshis E, Aiello AE, Galea S. Gender differences in the genetic and environmental determinants of adolescent depression. Depress. Anxiety. 2010;27:658–666. doi: 10.1002/da.20692. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Wight RG, Aneshensel CS, Botticello AL, Sepúlveda JE. A multilevel analysis of ethnic variation in depressive symptoms among adolescents in the United States. Soc. Sci. Med. 2005;60:2073–2084. doi: 10.1016/j.socscimed.2004.08.065. [DOI] [PubMed] [Google Scholar]

RESOURCES