Ethnic Differences in Clinical Presentation of Depression in Adult Women
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
R E S E A R C H R E P O R T
Ethnic Differences in Clinical Presentation
of Depression in Adult Women
HECTOR F. MYERS
University of California, Los Angeles, and Charles R. Drew University of
Medicine and Science
IRA LESSER
Harbor–UCLA Medical Center
NORMA RODRIGUEZ
Pitzer College
CONSUELO BINGHAM MIRA
Charles R. Drew University of Medicine and Science
WEI-CHIN HWANG
University of California, Los Angeles
CHRISTINA CAMP
DORA ANDERSON
Harbor–UCLA Medical Center
LUCY ERICKSON
Charles R. Drew University of Medicine and Science
MARCY WOHL
Harbor–UCLA Medical Center
• Hector F. Myers, Department of Psychology, University of California, Los Angeles (UCLA), and
Department of Psychiatry, Charles R. Drew University of Medicine and Science; Ira Lesser,
Christina Camp, Dora Anderson, and Marcy Wohl, Department of Psychiatry, Harbor–UCLA
Medical Center; Norma Rodriguez, Department of Psychology, Pitzer College; Consuelo Bingham
Mira and Lucy Erickson, Department of Psychiatry, Charles R. Drew University of Medicine and
Science; Wei-Chin Hwang, Department of Psychology, UCLA.
Christina Camp is now at the California School of Professional Psychology.
This research is supported in part by National Institute of Mental Health Grant MH47913 to
the Research Center on the Psychobiology of Ethnicity, Harbor–UCLA Medical Center, and by Na-
tional Institutes of Health (NIH) Grant 1 P20 RR11145 to the Research Centers in Minority
Institutions Clinical Research Center, Charles R. Drew University of Medicine and Science, and
NIH Grant RR00425 to the NIH General Clinical Research Center, Harbor–UCLA Medical Center.
Correspondence concerning this article should be addressed to Hector F. Myers, Department of
Psychology, UCLA, P. O. Box 951563, Los Angeles, California 90095-1563. E-mail:
myers@psych.ucla.edu
Cultural Diversity and Ethnic Minority Psychology Copyright 2002 by the Educational Publishing Foundation
Vol. 8, No. 2, 138–156 1099-9809/02/$5.00 DOI: 10.1037//1099-9809.8.2.138
138ETHNIC DIFFERENCES IN DEPRESSION 139
This study examined ethnic differences in self-report and interviewer-rated depressive
symptoms and estimated the contributions of sociodemographic and psychosocial factors
in predicting severity of depression. One hundred twenty-five clinically depressed Afri-
can American (n = 46), Caucasian (n = 36), and Latina (n = 43) women were re-
cruited. After controlling for differences in socioeconomic status, African American
women reported more symptoms of distress and Latinas were rated as significantly more
depressed than the other groups. However, these ethnic differences were not moderated by
either education or employment. Finally, hierarchical regression analysis indicated that
severity of depression was predicted by low education, being single, being Latina, high
perceived stress, and feelings of hopelessness. Additional research is needed to validate
these results and to investigate their clinical significance.
There has been increasing interest in the 2000; Merikangas, 2000). One of the most
role ethnicity and culture play in psychiatric consistent findings in this literature is the
illness, especially in depression (Lu, Lim, & greater prevalence of major depression
Mezzich, 1995). However, relatively few among women in the United States and
studies have systematically investigated eth- other developed countries, with most stud-
nic differences in the experience and ex- ies reporting a 2:1 female-to-male ratio (Cul-
pression of depression and related psycho- bertson, 1997; Kessler, 2000). Various expla-
logical distress or the implications that such nations have been offered to account for
differences might have for diagnosis, treat- this gender difference, including socioeco-
ment, and treatment outcome (Lawson, nomic, biological, personality, differential
Hepler, Holladay, & Cuffle, 1994; Myers, exposure to chronic stresses, cognitive and
1993; Zhang & Snowden, 1999). As the U.S. coping styles, and differential patterns of ex-
society becomes progressively more multi- pression of distress (Abramson et al., 1999;
ethnic, there is an increased need for well- McGrath, Keita, Strickland, & Russo, 1990;
designed studies that can help improve di- Nolen-Hoeksema, 1995). However, fewer
agnostic accuracy, quality, and effectiveness studies have investigated the role that eth-
of treatment and prevention services for nicity might play in accounting for differ-
clinically depressed patients from diverse ences in the experience and expression of
backgrounds (Neighbors et al., 1992). Thus, depression in general, and specifically in
the present study investigated possible eth- women.
nic differences in reported depressive symp- All of the evidence to date indicates that
toms, as well as in the relative contributions although there are ethnic differences in the
of socioeconomic status (SES), sociodemo- expression of depressive symptoms across
graphic, and psychosocial risk and protec- ethnic groups, these differences are rela-
tive factors in predicting severity of depres- tively minor, and that there is remarkable
sion in a multiethnic community sample of consistency in the core features of depres-
clinically depressed women. sion across countries and ethnic groups
In the United States and worldwide, de- (Ballenger et al., 2001; Weissman et al.,
pressive disorders account for a significant 1996). Nevertheless, these studies have re-
proportion of psychiatric disorders, with ported a greater tendency among depressed
women, the poor, young adults, the unem- Caucasians and African Americans to report
ployed, the unmarried, and Latinos experi- cognitive-affective symptoms, with African
encing a disproportionate burden of mor- Americans reporting more anxiety, anger,
bidity (Blazer, Kessler, McGonagle, & and hostility than Caucasians (Fabrega, Mez-
Swartz, 1994; Culbertson, 1997; Kessler, zich, & Ulrich, 1988; Raskin, Crook, & Her-140 MYERS ET AL .
man, 1975). There is also evidence suggest- support systems, and psychiatric risk factors
ing that African Americans are more likely such as personal and family psychiatric and
to report suspiciousness or paranoia than medical history in risk for depression (Got-
Caucasians and that this can affect the qual- lib & Hammen, 1992; Kessler & Magee,
ity of counselor–client relationship and like- 1993). However, relatively little attention
lihood of early treatment termination (Ter- has been given to testing the contribution of
rell & Terrell, 1984; Watkins & Terrell, these factors in conferring risk for depres-
1988; Whaley, 1998). However, most of this sion in people of color. Nevertheless, some
latter work has been conducted on college research indicates that people of color, es-
students and on patients suffering from pecially women of color, face a dispropor-
schizophrenic disorders rather than on pa- tionate burden of psychosocial risk from a
tients with depression. variety of sources, including chronic and
Somewhat less consistent is evidence sug- episodic life events, SES, acculturation, eth-
gesting that African Americans and Latinos nic/minority pressures, family, and environ-
are also more likely to report somatic com- mental stresses, and that these risks are as-
plaints than Caucasians (Compton & Jones, sociated with a disproportionate burden of
1991; Escobar, Rubio-Stipec, Canino, & psychiatric morbidity and mortality (Chis-
Karno, 1989; Kirmayer & Young, 1998; holm, 1996; Clark, Anderson, Clark, & Wil-
Roberts, 1992; Wohl, Lesser, & Smith, liams, 1999; Crittle, 1996; Hovey, 2000;
1997). This tendency to somatize distress is Rogler, Cortes, & Malgady, 1991; Roysircar-
generally associated with lower SES, female Sodowsky & Maestas, 2000; Salgado de
gender, older age, monolingual Spanish Snyder, 1987; D. R. Williams, Yu, Jackson, &
speakers, immigrants, and low-acculturated Anderson, 1997). However, little is known
individuals (Escobar et al., 1989; Guernac- about the contribution that psychosocial
cia, Angel, & Worobey, 1989; Kolody, Vega, and psychiatric risk factors make in account-
Meinhart, & Bensussen, 1986; Noh, Avison, ing for severity of depression or in shaping
& Kaspar, 1992; Swenson, Baxter, Shetterly, the expression of depression in women of
Scarbro, & Hamman, 2000). color (Myers, 1993; Nolen-Hoeksema,
It is not clear what accounts for these 1995).
group differences, although cultural factors Studies have also shown that social un-
have been implicated (Ballenger et al., 2001; dermining and low social support are re-
Weissman et al., 1996). It is also not clear lated to depression and psychological dis-
what effect these differences in the expres- tress in African American women (Gant et
sion of depressive symptoms might have on al., 1993) and in Latina immigrants (Sal-
diagnostic decisions and on estimates of dis- gado de Snyder, 1987). However, studies ex-
ease prevalence. However, Neighbors and amining the role of social support as a mod-
colleagues (Neighbors, Jackson, Campbell, erator of the effects of stress on distress have
& Williams, 1989), Whaley (1998), and oth- yielded mixed results. Some studies confirm
ers have argued that such differences in that social support moderates the stress–
symptom expression, however minor, are distress relationship in Latino and African
one factor that might account for the appar- American adults and youths (Leadbeater &
ent greater risk of psychiatric misdiagnosis Linares, 1992; Padilla, Cervantes, Mal-
experienced by African Americans and donado, & Garcia, 1988; Warren, 1997),
might complicate treatment decisions (Law- whereas other studies fail to confirm this re-
son, 1996). lationship (Crittle, 1996; Snapp, 1992).
There is also extensive evidence of the These findings suggest a need to investigate
complex interplay of psychosocial risk fac- the interplay of social undermining, stress,
tors such as stress and cognitive appraisal and social support in depression among
associated with chronic and episodic life women from different ethnic groups.
events, childhood adversity, deficits in social Although there is suggestive evidence ofETHNIC DIFFERENCES IN DEPRESSION 141
ethnic differences in the experience and ex- can American and Latina women would re-
pression of depression, this research is of port more depressive symptoms and be
mixed quality, often fails to distinguish be- rated as more severely depressed than Cau-
tween minor syndromal differences and dif- casian women; (b) that the pattern of symp-
ferences in core features of the disorder, of- toms would differ such that compared with
ten fails to control for demographic or other Caucasian women, African American
potentially confounding variables, and relies women would report more hostility, suspi-
exclusively on self-report symptom measures ciousness, and somatic complaints (Raskin
in nonclinical samples (Myers, 1993). Fur- et al., 1975; Whaley, 1998) and Latinas
thermore, it is unclear whether the observed would report more phobic anxiety and so-
ethnic differences in symptom expression matic complaints (Aneshensel et al., 1983;
are attributable to social class. For instance, Roberts, 1992); and (c) that some of these
research by Aneshensel, Clark, and Frerichs differences are expected to be evident even
(1983) found greater depressive symptoms after controlling for SES. However, we also
among Latinos in comparison with Cauca- expect that SES will moderate the relation-
sians and African Americans; however, once ships between ethnicity and severity of de-
SES was controlled, the group differences pression, such that the greatest differences
disappeared. It is also unclear whether social are observed among the most economically
class serves as a moderator, such that within disadvantaged women. Finally, and consis-
a given ethnic group, the greatest differ- tent with prevailing evidence, we hypoth-
ences in distress and depression are ob- esized (d) that SES, psychosocial, and psy-
served among the lower SES, but few or no chiatric risk and protective factors will
differences are found among the more af- predict severity of depression independent
fluent. It is also of interest to pursue this of ethnicity.
moderation effect among ethnically diverse,
yet economically similar, populations to in-
vestigate whether ethnicity interacts with Method
SES under these circumstances (Betancourt
& Lopez, 1993). Therefore, research is still Participants
needed to investigate whether there are
stable and consistent ethnic differences in Clinically depressed African American, Cau-
reported symptoms of depression and in se- casian, and Latina women between the ages
verity of the disorder in clinically depressed of 18–65 years who sought treatment at one
people. of two public psychiatric outpatient clinics
The purpose of this study, therefore, was were recruited to participate in this study of
to investigate ethnic differences in self- ethnic differences in clinical depression. A
reported and interviewer-rated depressive total of 461 women were identified as poten-
symptoms, to explore the role SES plays in tially eligible and were screened by tele-
explaining such differences, and to estimate phone to determine if they met the follow-
the relative contributions of sociodemo- ing study criteria: self-identified as African
graphic and psychosocial factors in predict- American, Caucasian, or Latina; had parents
ing severity of depression. These questions and grandparents of the same ethnic group;
were investigated in a sample of depressed sought treatment for current depression;
adult African American, Caucasian, and had no history of mania or psychosis; and
Latina women who sought treatment at one reported no alcohol or illicit drug abuse in
of two Los Angeles public mental health the past year. One-hundred and thirty-six
clinics. We hypothesized, consistent with women met these screening criteria and
some previous studies (Brown, Schulberg, & were invited for a formal assessment inter-
Madonia, 1996; Malgady & Rogler, 1993; view with a trained, ethnically and linguisti-
Raskin et al., 1975), (a) that depressed Afri- cally matched interviewer. Eleven partici-142 MYERS ET AL .
pants were excluded because they met ported a history of psychiatric disorders,
criteria for Bipolar I or II, evidenced current 58% (n = 73) reported a family history of
psychotic symptoms or had a history of psy- psychiatric disorders, and 50% (n = 62) re-
chosis, met criteria for drug or alcohol ported a history of other medical problems.
abuse or dependence, or had a physical ill- However, there were no ethnic differences
ness that might account for their depression in psychiatric or medical history.
(e.g., severe anemia, thyroid dysfunction). Given the extensive amount of time re-
Thus, the final sample consisted of 125 quired for the interview and the fact that
women (46 African American, 36 Caucasian, data were collected entirely for research
and 43 Latina). purposes with no specific benefits to the
As seen in Table 1, participants across participants, participants were paid $75
the three ethnic groups were similar in age for the interview and $25 for lab samples, for
(M = 42.03 years, SD = 11.03) and reported a total of $100. All participants were treated
comparably low annual incomes (M income in accordance with the American Psycho-
= $7,934/annum, SD = $5,942). However, logical Association ethical guidelines and
Latinas reported significantly lower educa- principles.
tion than African Americans and Cauca-
sians, F(2, 122) = 45.98, p = .000, and were
Measures
primarily non-U.S. born (84%, n = 36). Al-
though the non-U.S. born Latinas reported DEMOGRAPHIC AND BACKGROUND CHARACTER-
an average of 22.8 years of residency in the ISTICS. Age, marital status, years of educa-
United States (range = 4–51 years), all spoke tion, occupational status and history,
Spanish as their primary language. Latinas monthly income, place of birth, number of
also reported consuming significantly less al- years in the United States (if non-U.S.
cohol per week, F(2, 122) = 3.51, p = .03, born), primary language, and English profi-
than the Caucasian women. ciency (if nonnative English speaker) were
As seen in Table 2, Latinas were also obtained from participants.
more likely to be employed, 2(2, N = 125) =
18.77, p = .000, although they were earning INTERVIEWER-RATED SYMPTOM SEVERITY MEA-
less income and were more likely to be mar- SURES. The Structured Clinical Interview for
ried/cohabiting, 2(2, N = 125) = 7.21, p = DSM–IV (SCID–IV; First, Spitzer, Gibbon, &
.03, than either the African American or Williams, 1997) was used to assess whether
Caucasian women. As a group, the women patients met DSM–IV criteria for major de-
reported a substantial burden of psychiatric pression or dysthymia. The majority of the
and medical morbidity; 71% (n = 89) re- women enrolled in the study met criteria for
TABLE 1 Means and Standard Deviations of Demographic Variables by Ethnic Group
African Americans Latinas Caucasians Total
(n = 46) (n = 43) (n = 36) (n = 125)
Variable M SD M SD M SD M SD
Age 43.67 10.84 41.77 11.41 40.25 10.81 42.03 11.03
Education 12.40a 1.82 7.71a, b 3.86 13.00b 2.11 10.96 3.62
Yearly incomea 7.74 5.89 7.88 4.78 9.10 7.12 7.93 5.94
No. of drinks weekly 0.22 0.59 0.09a 0.37 0.39a 0.05 0.23 0.51
No. of cigarettes daily 3.91 7.08 2.48 4.95 3.04 5.27 3.14 5.77
Note. Means in the same row that share the same subscripts differ significantly at p < .05.
a
Yearly income is reported in thousands.ETHNIC DIFFERENCES IN DEPRESSION 143
TABLE 2 Percentages and Sample Sizes of Demographic Variables by Ethnic Group
African
Americans Latinas Caucasians Total
(n = 46) (n = 43) (n = 36) (n = 125)
Variable % n % n % n % n
% employed 19.6a 9 62.8a, b 27 30.6b 11 37.6 47
% married/cohabiting 10.9a 5 30.2a, b 13 11.1b 4 17.6 22
% using recreational drugs 4.3 2 4.7 2 11.1 4 6.4 8
% with psychiatric history 71.7 33 69.8 30 72.2 26 71.2 89
% with family psychiatric history 69.6 32 46.5 20 58.3 21 58.4 73
% with history of medical illnesses 56.5 26 46.5 20 44.4 16 49.6 62
Note. Percentages in the same row that share the same subscripts differ significantly at p < .05.
major depression (n = 120), and 5 met cri- report symptom measure that uses a 5-point
teria for dysthymia. No differences between scale of distress ranging from 0 = not at all to
these groups were noted on any of the vari- 4 = extremely. The BSI has been shown to be
ables of interest, therefore the analyses were reliable and valid with a variety of popula-
run on the entire sample. tions (Hemmings et al., 1999; Skilbeck et al.,
Severity of depressive symptoms was 1985). The BSI yields three global scores,
evaluated using the structured interview but for the purposes of this study, only the
guide for the Hamilton Rating Scale for De- Global Severity Index (GSI), which is the
pression (HRSD-24; Hamilton, 1960; J. B. W. sum of all items responded to (␣ = .94), was
Williams, 1988). This scale includes 24 symp- used as a measure of overall level of psycho-
toms of depressive illness rated on either a logical distress. The BSI also yields nine
3-point or 5-point scale that ranges from 0 = symptom subscale scores: somatization, ob-
absent to 2 or 4 = severe (␣ = .85; = .78). Only sessive–compulsive, interpersonal sensitivity,
patients whose depression was rated as at depression, anxiety, hostility, phobic anxi-
least of mild severity (i.e., HRSD > 11) were ety, paranoid ideation, and psychoticism.
included in the study. However, to avoid redundancy with the BDI
and to reduce testing burden, we omitted
SELF-REPORT SYMPTOM SEVERITY MEASURES. the items on the depression subscale from
Depressive symptoms were assessed using the subscale comparisons. Subscale scores of
the Beck Depression Inventory (BDI; Beck, the BSI were found to have moderate to
1972), a 21-item self-report measure that as- high reliability in the present sample (al-
sesses cognitive, affective, and somatic symp- phas ranged from .59 to .85).
toms of depression. Participants were asked
to rate the degree of severity of each symp- LAB TESTS. Blood and urine samples were
tom on a 4-point scale from 0 = no symptoms taken and assayed to assess participants’
to 3 = severe symptoms. Consistent with previ- overall health and nutritional status and cur-
ous research (Hemmings, Reimann, Madri- rent substance use. Lab results were made
gal, & Velasquez, 1999; Skilbeck, Acosta, Ya- available to participants, and those with
mamoto, & Evans, 1985), the BDI evidenced acute and untreated conditions (e.g., thy-
good internal consistency in this sample roid dysfunction, anemia) were excluded
(␣ = .83). from the study and referred for treatment.
Psychological distress was assessed using
the Brief Symptom Inventory (BSI; Dero- PSYCHIATRIC RISK FACTORS. Participants were
gatis & Melisaratos, 1983), a 53-item self- queried about their history of use of alcohol144 MYERS ET AL .
and illicit drugs, whether they had ever been (CSQ; Abramson, Metalsky & Alloy, 1993)
diagnosed or received treatment for any psy- was used to assess feelings of hopelessness
chiatric disorders, whether any member of and helplessness, which is conceptualized as
their immediate family had ever been diag- a cognitive style risk factor for depression
nosed or treated for a psychiatric disorder, (Abramson et al., 1999; Alloy, Kelly, Mineka,
and their history of medical problems. The & Clements, 1990). The CSQ is a 20-item
number of drinks in an average week was true-false scale that measures negative ex-
computed, as well as the number of ciga- pectations and other dysfunctional cogni-
rettes smoked daily. Affirmative responses tions (e.g., “I just don’t get breaks, and
on drug use, on personal and family psychi- there’s no reason to believe I will in the fu-
atric history, and on medical history were ture” and “My future seems dark to me”).
scored 1 and negative responses 0, and sepa- Items were recoded, scored, and summed
rate scores were computed for each variable. such that a low score indicates high feelings
of hopelessness. This sum score had good
PSYCHOSOCIAL RISK FACTORS. Chronic stress reliability (␣ = .89).
was assessed with a 21-item revised self- A short three-item Social Undermining
report Role Strain Questionnaire (Myers, Scale (SUS; Vinokur & Van Ryn, 1993) was
1985) that assesses on a 4-point scale (1 = not administered to assess stresses that are
a problem to 4 = a major problem) the amount caused by members of each participant’s so-
of difficulty participants experienced in cial network as an index of interpersonal
the past month from a number of sources, stress. Participants were asked to rate, on a
including economic, employment, crime, 5-point rating scale (1 = not at all to 5 = a
legal problems, discrimination, housing, great deal), the degree to which four impor-
transportation, child care, personal con- tant people in their network “act angry or
flicts, and illness or accidents. A total unpleasant,” “criticize,” or “make life diffi-
chronic stress score was calculated by sum- cult for you.” Responses were summed to
ming across all items responded to. Al- form an interpersonal stress score, with
though this score was only modestly reliable higher scores reflecting higher levels of in-
(␣ = .56), this is most likely attributable to terpersonal stress. The SUS was found to be
the fact that strains in one domain (e.g., moderately reliable (␣ = .77).
family conflict) need not be related to Finally, social support, which is concep-
stresses in other domains (e.g., living in a tualized as a protective resource that may
high crime area), yet an overall chronic moderate stress to reduce the severity of de-
strain score provides a meaningful index of pression or facilitate recovery (Brugha et al.,
differences in stress burden that each par- 1987; Brugha et al., 1990; McLeod, Kessler,
ticipant carries. & Landis, 1992), was assessed with a short
Perceived stress was assessed with the five-item version of the Social Support Ques-
14-item Perceived Stress Scale (Cohen, Kar- tionnaire (Sarason, Levine, Basham, & Sara-
marck & Mermelstein, 1983) that measures son, 1983). The measure asks participants to
the degree to which life events are evaluated identify and rate the four most important
as uncontrollable, emotionally overloading, people in their lives on the degree to which
and unpredictable (e.g., “In the last month, they provide advice, emotional support, and
how often have you felt nervous or stressed” instrumental support. Ratings are made on
and “In the last month, how often have you a 5-point scale from 1 = not at all to 5 = a
found that you could not cope with all the great deal. An overall sum score was calcu-
things you had to do”). Respondents rated lated (␣ = .91), with higher scores reflecting
each item on a 5-point scale from 0 = never to higher levels of support.
4 = very often, and a sum score was calculated
that was moderately reliable in this sample Procedure
(␣ = .73). At the interview, trained and ethnically and
The Cognitive Style Questionnaire linguistically matched female interviewersETHNIC DIFFERENCES IN DEPRESSION 145
described the purpose of the study and An a priori alpha level of .05 was used for all
study procedures and requirements, written of the statistical analyses reported.
consent to participate was obtained, biologi- Because Latinas were significantly less
cal samples were collected, the diagnostic educated but were more likely to be em-
interview and interviewer-rated assessments ployed and married or cohabiting than Af-
were conducted, and self-report measures rican American and Caucasian women,
were administered. All interviewers were these variables were treated as covariates in
trained and supervised in the administration subsequent analyses. These covariates were
of the diagnostic interviews and in rating the dichotomized into the following values: edu-
range and severity of depressive symptoms cation: 12 years
by one of the coauthors (Ira Lesser), with of education; employment status: full/part
periodic reviews of diagnostic procedures time or unemployed; and marital status:
and consultation on the more difficult cases. married/cohabiting or single/unattached.
All of the instruments were originally de- Although Latinas also consumed signifi-
veloped in English and subsequently trans- cantly less alcohol than the other women,
lated into Spanish by a certified translator. the quantities of alcohol were so small and
The Spanish version of each instrument was unlikely to have a significant role in depres-
then back-translated by an English/Spanish sion that this variable was omitted from sub-
bilingual individual familiar with psychologi- sequent analyses to save degrees of freedom.
cal research, and discrepancies between the
English and Spanish versions of the protocol
were resolved, thus “decentering” the origi-
nal English version (Brislin, 1993). Thirty- Differences in Clinical Status
six (84%) of the Latina women chose to be
interviewed in Spanish, and the remaining 7 A MANCOVA controlling for education,
(16%) were interviewed in English. Both the employment, and marital status was run test-
English and Spanish protocols were pre- ing for group differences on severity of de-
tested on a subsample of patients and non- pression as measured by the HRSD-24, de-
patients, and minor additional changes were pressive symptoms as measured by the BDI,
made to improve the clarity and compre- and psychological distress as measured by
hensibility of the measures. the BSI–GSI.
As shown in Table 3, there was a signifi-
cant main effect of ethnicity on clinical sta-
tus, Wilks’s lambda F(6, 234) = 4.74, p = .000.
Results
Between-subjects tests indicated that there
were significant group differences on inter-
A series of univariate analyses tested for eth- viewer-rated severity of depression (HRSD-
nic group differences at entry into treat- 24), F(2, 119) = 10.69, p = .000, and on self-
ment on demographic characteristics. Sub- reports of psychological distress (GSI),
sequently, we used a series of multivariate F(2,119) = 3.04, p = .05. Pairwise compari-
analyses of covariance (MANCOVA), con- sons indicated that Latinas were rated as
trolling for education, employment, and more severely depressed than both African
marital status to test for ethnic differences in American (p = .000) and Caucasian (p =
clinical status, symptom expression, and psy- .001) women and that African American
chosocial factors. Next, we conducted a hi- women reported significantly more psycho-
erarchical regression analysis to investigate logical distress than Caucasian women (p =
whether SES moderates the observed ethnic .05) but not more than Latinas. There was
differences in severity of depression. Finally, also a strong trend (p = .06) on the BDI, with
we ran a hierarchical regression analysis to Latinas reporting more symptoms of depres-
identify which of the factors studied were sion than Caucasian women (p = .05) but
the best predictors of severity of depression. not more than African American women.146 MYERS ET AL .
TABLE 3 Adjusted Means and Standard Errors of Clinical Status at Entry Into Treatment by
Ethnic Group
African
Americans Latinas Caucasians Total
(n = 46) (n = 43) (n = 36) (n = 125)
Variable M SE M SE M SE M SE
Interviewer-rated
HRSD 23.34a 1.31 32.79a, b 1.50 24.40b 1.51 26.84 .77
Self-report
BDI total 31.01 1.37 34.12 1.57 28.35 1.58 31.16 .80
BSI Global Score GSI 58.73a 3.71 56.36 4.26 45.45a 4.30 53.51 2.18
Note. Means that share the same subscript differ at p < .05. HRSD = Hamilton Rating Scale for Depression; BDI = Beck Depression
Inventory; BSI = Brief Symptom Inventory; GSI = Global Severity Index.
Differences in Symptom Expression F(2, 119) = 4.54, p = .01. Pairwise compari-
sons indicated that African American
To test the hypothesis of ethnic differences women reported more paranoid symptoms
in symptom expression, we included all of (p = .001) than Caucasian women, and both
the BSI subscales except the depression sub- African American women (p = .02) and Lati-
scale in a MANCOVA controlling for educa- nas (p = .05) reported significantly more so-
tion, employment, and marital status. As matic symptoms than Caucasian women.
shown in Table 4, a significant overall differ- Strong trends (p = .06) were also observed
ence was obtained, Wilks’s lambda F(16, on paranoid symptoms and anxiety, with
224) = 2.32, p = .003; however, only partial Latinas reporting more of these symptoms
support for this hypothesis was obtained. than Caucasian women. However, the ex-
More specifically, between-subjects tests in- pected differences on symptoms of hostil-
dicated that significant differences were ity (African American > Caucasian) and
only apparent on paranoid symptoms, F(2, phobic anxiety (Latina > Caucasian) were
119) = 7.57, p = .001, and on somatization, not observed.
TABLE 4 Adjusted Means and Standard Errors of Symptom Expression of Depression at
Treatment Entry by Ethnic Group
African
Americans Latinas Caucasians Total
(n = 46) (n = 43) (n = 36) (n = 125)
Brief Symptom Inventory M SE M SE M SE M SE
Obsessive–compulsive 9.37 .73 9.73 .83 8.77 .84 9.29 .43
Interpersonal sensitivity 5.27 .49 5.75 .57 4.80 .57 5.42 .29
Anxiety 7.32 .64 8.93 .73 6.28 .74 7.51 .37
Hostility 4.93 .54 3.26 .62 4.01 .62 4.06 .32
Phobic anxiety 5.51 .64 5.34 .74 4.09 .75 4.98 .38
Paranoid ideation 6.20a .49 5.40 .56 3.38a .57 4.99 .29
Psychotic 5.75 .46 4.59 .53 4.25 .54 4.86 .27
Somatization 7.10a .67 7.15b .78 4.31a, b .77 6.19 .43
Note. Means that share same subscript differ at p < .05.ETHNIC DIFFERENCES IN DEPRESSION 147
Differences on Psychosocial Factors moderates the observed ethnic differences
in severity of depression (HRSD-24).
A MANCOVA controlling for years of edu- In each analysis, education and employ-
cation, employment, and marital status were ment served as the indicators of SES, and
run testing for ethnic differences on feelings following the procedures suggested by
of hopelessness, chronic stress, perceived Baron and Kenny (1986), variables were en-
stress, social support, and social undermin- tered in sets and in three steps. In Step 1,
ing. As shown in Table 5, results indicated education and employment were entered.
that there was a significant overall ethnic dif- In Step 2, ethnicity was dummy coded into
ference, Wilks’s lambda F(10, 230) = 1.85, Ethnic 1 (African American compared with
p = .05; however, between-subjects tests indi- Caucasian) and Ethnic 2 (African American
cated that only one significant group differ- compared with Latina) and entered. African
ence on social support was obtained, F(2, Americans were chosen as the referent
119) = 3.01, p = .05. African American group because of its larger sample size,
women reported receiving more social sup- which would yield greater stability in the re-
port than Caucasian women. There was also sults. In Step 3, four interaction terms were
a trend on social undermining (p = .08), entered: Education × Ethnic 1, Education ×
with Latinas reporting less social undermin- Ethnic 2, Employment × Ethnic 1, and Em-
ing than Caucasian women. No differences ployment × Ethnic 2.
on feelings of hopelessness, chronic stress,
or perceived stress were obtained.
Moderation of Ethnicity on Severity of
Depression by SES
SES as a Moderator of Ethnic Differences
As seen in Table 6, the regression equation
Studies of ethnic differences in mental testing SES as a moderator on severity of
health often fail to adequately account for depression accounted for 20% of the vari-
the possible effects of SES. It is very likely, ance, F(8, 116) = 4.75, p = .000. In Step 1,
for example, that SES may exert its effects education was significantly and negatively
on clinical outcomes either independent of related to the severity of depression, ac-
ethnicity or as a moderator of ethnicity. In counting for 16% of the total variance ex-
the previous analyses, the independent ef- plained, Fchange(2, 122) = 11.84, p = .000.
fects of SES were statistically controlled be- This effect indicated that low education con-
fore testing for ethnic differences. In the tributes significantly to the severity of de-
next analyses, we investigate whether SES pression. In Step 2, although the step was
TABLE 5 Adjusted Means and Standard Errors of Psychosocial Factors at Entry Into Treatment by
Ethnic Group
African
Americans Latinas Caucasians Total
(n = 46) (n = 43) (n = 36) (n = 125)
Variable M SE M SE M SE M SE
Hopelessness 28.32 0.89 26.81 1.02 28.64 1.03 27.92 0.52
Chronic stress 41.70 1.05 39.86 1.20 39.78 1.21 40.44 0.61
Perceived stress 37.42 0.97 39.86 1.11 38.64 1.12 38.64 0.57
Social support 74.86a 2.43 70.24 2.79 66.10a 2.81 70.40 1.43
Social undermining 23.46 1.18 20.93 1.35 25.63 1.36 23.34 0.69
Note. Means that share the same subscript differ at p < .05.148 MYERS ET AL .
TABLE 6 Hierarchical Regression Analysis of Moderation of Ethnicity on Hamilton Rating Scale
for Depression by Socioeconomic Status (N = 125)
Step Variable B SE B  R2 change F change
1 Socioeconomic status 0.16 11.84***
Education −3.80 0.82 −0.41***
Employment −0.35 1.68 −0.02
2 Ethnicity 0.04 2.78
Ethnic 1 1.68 1.90 0.08
Ethnic 2 5.38 2.29 0.28*
3 Education & Employment × Ethnicity Interaction Terms 0.05 1.81
Education × Ethnic 1 −1.26 5.80 −0.14
Education × Ethnic 2 −4.95 3.32 −0.53
Employment × Ethnic 1 0.44 1.24 0.05
Employment × Ethnic 2 2.89 1.71 0.31
Note. R2 = 0.25 (adjusted R2 = 0.20), F(8, 116) = 4.75, p = .000. Ethnic 1 was coded 0 = African American, 1 = Caucasian, and Ethnic
2 was coded 0 = African American, 1 = Latina.
*p < .05. ***p < .001.
not significant, ethnicity made a marginal moderates stress in predicting severity of
additional contribution, Fchange(2, 120) = depression.
2.78, p = .07, indicating that being Latina as As shown in Table 7, the regression
opposed to African American predicted equation predicting severity of depression
more severe depression. In Step 3, none of accounted for 32% of the variance, F(11,
the interaction terms were significant, which 113) = 6.21, p = .000. In Step 1, low educa-
indicates that neither education nor em- tion and being single were significant inde-
ployment moderated ethnicity in account- pendent predictors of severity of depression,
ing for differences on severity of depression. Fchange = 9.43, p = .000, and the entire step
accounted for 19% of the total variance. In
Step 2, ethnicity was significantly related to
Predictors of Severity of Depression
depression and accounted for an additional
We conducted a hierarchical regression 5% of the variance, Fchange = 3.70, p = .03.
analysis to determine which psychosocial This effect indicated that after controlling
factors predict severity of depression as in- for differences in demographic characteris-
dexed by HRSD-24 scores and controlling tics, Latinas were rated as more severely de-
for SES, ethnicity, and psychiatric risk fac- pressed than African Americans. Personal
tors. Variables were entered in sets and in and family psychiatric history in Step 3 was
seven steps: Education, marital status, and not significant. In Step 4, perceived stress
employment were entered in Step 1; ethnic- was positively and significantly associated
ity was entered in Step 2 as dummy variables with severity of depression, Fchange = 11.10,
Ethnic 1 (African American compared with p = .001, and accounted for an additional
Caucasian) and Ethnic 2 (African American 6% of the variance. In Step 5, after control-
compared with Latina); personal and family ling for all previous steps, hopelessness
psychiatric history were entered in Step 3; accounted for an additional 7% of
perceived stress was entered in Step 4; hope- the variance, Fchange = 4.77, p = .03, indicat-
lessness as an index of psychological vulner- ing that the greater the feelings of hope-
ability was entered in Step 5; social support lessness, the greater the severity of de-
was entered in Step 6; and a Perceived Stress pression. Contrary to expectations, however,
× Social Support interaction term was en- social support in Step 6 and the Perceived
tered in Step 7 to test whether social support Stress × Social Support interaction term inETHNIC DIFFERENCES IN DEPRESSION 149
TABLE 7 Hierarchical Regression Analysis of Severity of Depression With Background
Characteristics, Ethnicity, Psychiatric History, Perceived Stress, Hopelessness, and Social Support
as Predictors (N = 125)
Step Variable B SE B  R2 change F change
1 Background characteristics 0.19 9.43***
Marital status −4.07 2.03 −0.17*
Employment 0.03 1.68 0.00
Education −4.01 0.82 −0.43***
2 Ethnicity 0.05 3.70*
Ethnic 1 1.66 1.87 0.08
Ethnic 2 6.16 2.26 0.32**
3 Psychiatric history 0.03 1.98
Participant psychiatric history −1.12 1.73 −0.06
Family psychiatric history 3.14 1.58 0.17*
4 Perceived stress 0.38 0.11 0.27 0.06 11.10**
5 Hopelessness −0.16 0.07 −0.20 0.03 4.77*
6 Social support −1.27 0.73 −0.14 0.02 2.97
7 Perceived Stress × Social Support interaction term 4.64 4.11 0.50 0.01 1.28
Note. R2 = 0.38 (adjusted R2 = 0.32), F(11, 113) = 6.21, p < .001. Ethnic 1 was coded 0 = African American, 1 = Caucasian and Ethnic
2 was coded 0 = African American, 1 = Latina.
*p < .05. **p < .01. ***p < .001.
Step 7 were unrelated to severity of depres- sian counterparts. It is also worth noting,
sion, which indicated that social support however, that no substantive differences in
had no direct protective effect and did not core symptoms of the disorder were ob-
moderate stress in predicting severity of served, which is consistent with previous
depression. findings of no differences in core features of
depression across ethnic or national groups
(Ballenger et al., 2001).
Discussion Although we hypothesized that African
American and Latina women would be rated
This study was conducted to investigate eth- as more severely depressed than Caucasian
nic differences in self-report and inter- women, our results indicated that only Lati-
viewer-rated depressive symptoms, to ex- nas were rated as more severely depressed,
plore the role SES plays in explaining such after controlling for differences in educa-
differences, and to estimate the relative con- tion, employment, and marital status. The
tributions of sociodemographic and psycho- greater severity ratings of Latinas are consis-
social factors in predicting severity of de- tent with previous evidence of more severe
pression in a multiethnic sample of clinically depression among less-educated, immi-
depressed adult women. Results from this grant, and monolingual Spanish speakers
study provided partial support for previous (Malgady & Rogler, 1993; Skilbeck et al.,
findings that ethnic groups differ in the ex- 1985; Swenson et al., 2000). We considered
pression of clinical depression assessed the possibility that this finding was an arti-
through self-report and interviewer-rated fact of measurement, but this is unlikely be-
measures (Ballenger et al., 2001; Fa- cause of the congruence between self-report
brega et al., 1988; Neff, 1984; Raskin et al., symptoms of distress and depression and the
1975; Wohl et al., 1997). As expected, Lati- interviewer ratings of symptom severity.
nas and African American women reported Thus, perhaps, the Latinas were more se-
more somatic complaints than their Cauca- verely depressed because of greater willing-150 MYERS ET AL .
ness to report symptoms of distress or be- differences in education, employment, and
cause of delay in seeking professional care marital status, only the comparison between
compared with the African American and the Latinas and Caucasian women ap-
Caucasian women. However, it is also pos- proached significance (p = .06). Neverthe-
sible that this difference could be due, at less, the self-report data in conjunction with
least in part, to a tendency of the clinical the interviewer ratings suggest that the Lati-
interviewer to rate the Latinas more se- nas, in particular, experience greater levels
verely. A report by Malgady and Costantino of depressive symptomatology compared
(1998) indicated that Latino clinicians rated with the other two groups. This higher level
psychiatric symptoms in bilingual Hispanics of depressive symptoms might be attributed
more severely than Caucasian clinicians, and to the additional sociocultural burdens
this was especially evident for bilingual and (e.g., acculturation) they face, as well as
monolingual Spanish interviewers. Unfortu- their postponing seeking professional men-
nately, this hypothesis could not be tested in tal health services. It is worth reinforcing,
this study because only trained, bilingual– however, that this is not a difference in the
bicultural Latinas interviewed the Latina core symptoms of depression but mainly a
patients, and thus, no comparisons by eth- difference in symptom reporting.
nicity of interviewer could be made. Never- Consistent with our expectations and
theless, these results await replication in controlling for differences in education, em-
future studies, and the competing hypoth- ployment, and marital status, depressed
eses of greater symptom reporting versus de- African American women also reported
layed treatment seeking versus ethnicity of greater psychological distress than Cauca-
the interviewer merit further investigation. sian women. These results confirm previous
Epidemiological studies indicate lower reports from community surveys (D. R.
rates of depression among recent Latino im- Brown & Gary, 1985; Dressler & Badger,
migrants compared with their U.S.-born 1985) and clinical studies (C. Brown et al.,
counterparts (Vega, Kolody, Aguilar- 1996; Raskin et al., 1975) that indicate a ten-
Graxiola, & Catalano, 1998). Unfortunately, dency for African American women to re-
because our Latina sample primarily con- port more psychological distress than Cau-
sisted of immigrants, we were not able to casian women. It is interesting to note,
compare them with their U.S.-born counter- however, that the African American women
parts. However, it is important to note that rated themselves as more psychologically
although low in acculturation (i.e., 84% distressed than the ratings given by the in-
were monolingual Spanish speakers), the terviewers would suggest. This apparent dis-
Latina women in our sample were not re- crepancy may reflect the fact that the range
cent immigrants, with an average of 23 years of symptoms available on the self-report
of residency in the United States. Thus, as measure is greater and more diverse than
reported in other studies, pressures of accul- the symptoms that are the focus of attention
turation, persistent socioeconomic disadvan- on the interviewer assessment of severity of
tage in a country of considerable wealth, ex- depression. Therefore, their overall level of
posure to anti-immigrant and racial distress may be more related to the fact that
prejudice, and other noxious social pro- they are slightly older, more likely to be un-
cesses may accumulate to erode whatever employed, less likely to be married or cohab-
protections these immigrants bring and con- iting, somewhat more likely to have a psychi-
tribute to increased psychiatric vulnerability atric history, and to carry a higher burden of
over time (Vega et al., 1998). medical morbidity than the Caucasian and
Although the African American and Latina women, rather than any substantive
Latina women reported more depressive difference in core symptoms of depression.
symptoms in the self-report measures than As shown in previous research (Terrell &
the Caucasian women, after controlling for Terrell, 1984; Whaley, 1998), our results alsoETHNIC DIFFERENCES IN DEPRESSION 151
indicated that African American women re- contrary to previous reports that found
ported more distrust than Caucasian women higher rates of somatic complaints among
and that these differences were not related Latinos (Compton & Jones, 1991; Golding,
to SES. This greater cultural mistrust on the Aneshensel, & Hough, 1991; Roberts, 1992;
part of African Americans has been shown Sylva, 1997). This finding was also somewhat
to influence their interactions with health unexpected given that the Latinas were sig-
and mental health care providers (Terrell & nificantly less educated and low accultur-
Terrell, 1984) and may increase the risk for ated. Nevertheless, the elevated levels of so-
clinicians to overinterpret the psychiatric matic complaints by the African American
significance of these symptoms rather than women may simply be reflective of their
seeing them as nonpathological idioms of overall greater psychological distress.
distress (Whaley, 1998). Contrary to expectations, there were no
Although not significant, Latinas also re- differences in hostility between African
ported more mistrust than Caucasian American women and the other women.
women, and along with the African Ameri- Previous studies reporting these differences
can women’s reported levels of mistrust were based on more severely depressed hos-
comparable with the adult psychiatric out- pitalized patients (Raskin et al., 1975) or on
patient sample used in the norming of the samples that included men (Fabrega et al.,
BSI (Derogatis & Melisaratos, 1983). The 1988) rather than on moderately depressed
Caucasian women, in contrast, reported lev- women. Similarly, the expected differences
els of mistrust more comparable with the on phobic anxiety between the Latina and
adult nonpatient norming sample of the the Caucasian women were not observed
BSI. It is reasonable to expect greater mis- (Karno et al., 1987). Additional research is
trust among residents in high-stress, high- needed to determine whether ethnic differ-
crime communities, such as those who par- ences in phobic anxiety are likely to be mod-
ticipated in this study. What is interesting, erated by severity of depression, gender, or
however, is that the same level of mistrust age.
was not evident in the Caucasian women in As suggested by several authors (Betan-
our sample. This difference in distrust is not court & Lopez , 1993; Myers, 1993; Williams
attributable to SES and suggests that it is et al., 1997), it is important to disentangle
likely to be culturally mediated, as suggested the effects of ethnicity from SES, so that ob-
by Watkins and Terrel (1988) and Whaley served group differences are not errone-
(1998). Additional exploration is warranted ously attributed to ethnicity. In the present
to examine whether higher levels of mistrust study, SES, as indexed by education, employ-
is a culturally congruent way of expressing ment, and marital status, either was statisti-
distress among women of color and whether cally controlled or its independent effects
lower social class exacerbates these feelings and its role as a possible moderator of eth-
of mistrust. nicity were tested. The results indicated that
The expected greater tendency toward there were ethnic differences in symptom
expressing somatic symptoms of distress reporting after controlling for differences
among women of color (C. Brown et al., on SES and that SES did not moderate eth-
1996; Escobar et al., 1989; Kolody et al., nicity in accounting for differences in sever-
1986; Noh et al., 1992) was confirmed, with ity of depression. Therefore, other sociocul-
both the African American and Latina tural factors that were not examined in
women reporting significantly more somatic these analyses may be implicated in account-
complaints than the Caucasian women, after ing for the group differences observed.
controlling for differences in SES. Although Finally, we also investigated what role
not statistically significant, the African psychosocial factors play in accounting for
American women reported more somatic differences in severity of depression. We ini-
complaints than the Latina women, which is tially investigated ethnic group differences152 MYERS ET AL .
in the psychosocial factors, controlling for depression cannot be determined from
SES. These initial comparisons yielded eth- these data. Subsequent analyses will explore
nic differences only on social support, with this question using independent ratings of
African American women reporting more chronic and episodic stress as predictors of
social support than Caucasian women. Nev- severity of depression.
ertheless, when these psychosocial factors Finally, these results are based only on
were entered into a regression equation, se- low-income and working-class women seek-
verity of depression was associated with feel- ing psychiatric care at public outpatient clin-
ings of hopelessness and higher perceived ics. It would be useful to determine whether
stress along with low education, being these results are replicated with comparable
single, and being Latina in comparison with samples of men and with men and women of
being African American. The significance of higher SES. It would also be useful to deter-
the strong association of feelings of hope- mine whether these minor syndromal differ-
lessness and perceived stress with depression ences observed at entry into treatment are
should not be overinterpreted because the associated with differences in diagnostic de-
tendency to overrate life difficulties is con- cisions or predict treatment received, treat-
founded with severity of depression (Gott- ment response, and treatment outcome over
lieb & Hammen, 1992), which precludes dis- time. All of the participants were in treat-
entangling the direction of the stress–de- ment, and subsequent articles will address
pression relationship. this issue.
The results also indicated that social sup-
port was not an independent predictor of
depression, nor did it serve as a moderator References
of stress. These findings are consistent with
previous studies that failed to support the Abramson, L. Y., Alloy, L. B., Hogan, M. E.,
hypothesized moderating effect of social Whitehouse, W. G., Donovan, P., Rose, D. J.,
support (Crittle, 1996; Snapp, 1992). How- et al. (1999). Cognitive vulnerability to de-
ever, these results are not entirely surprising pression: Theory and evidence. Journal of Cog-
because all of the women are clinically de- nitive Psychotherapy, 13, 5–20.
pressed. Therefore, whatever protective or Abramsom, L. Y., Metalsky, G. I., & Alloy, L. B.
moderating effect their social resources (1993). Hopelessness. In C. G. Costello (Ed.),
might have provided has already proven Symptoms of depression (pp. 181–205). New
inadequate. York: Wiley.
In summary, the results of this study pro- Alloy, L. B., Kelly, K. A., Mineka, S., & Clements,
vide additional, although somewhat limited C. M. (1990). Comorbidity of anxiety and de-
support, of ethnic differences in the expres- pressive disorders: A helplessness–hopeless-
sion of depression among clinically de- ness perspective. In J. D. Maser & C. R. Clon-
pressed women but not in the core symp- inger (Eds.), Comorbidity of mood and anxiety
disorders (pp. 499–543). Washington, DC:
toms of the disorder. These differences are
American Psychiatric Press.
not attributable to differences in SES and
Aneshensel, C. S., Clark, V. A., & Frerichs, R. R.
need to be replicated in larger and more
(1983). Race, ethnicity and depression: A
socioeconomically diverse samples. The re-
confirmatory analysis. Journal of Personality
sults also suggest the need to explore pos- and Social Psychology, 44, 385–398.
sible ethnocultural and psychosocial expla-
Ballenger, J. C., Davidson, J. R., Lecrubier, Y.,
nations for these differences. Our results Nutt, D. J., Kirmayer, L. J., Lepine, J. P., et al.
also confirm that depressed women, irre- (2001). Consensus statement on transcultural
spective of their ethnic background, experi- issues in depression and anxiety from the In-
ence high chronic strain. However, whether ternational Consensus Group on Depression
this greater stress burden is a contributor to and Anxiety. Journal of Clinical Psychiatry,
their depression or a consequence of their 62(Suppl. 13), 47–55.ETHNIC DIFFERENCES IN DEPRESSION 153
Baron, R. M., & Kenny, D. A. (1986). The mod- Journal of Health and Social Behaviors, 24,
erator–mediator variable distinction in social 385–396.
psychological research: Conceptual, strategic, Compton, A. B., & Jones, F. D. (1991). Clinical
and statistical considerations. Journal of Per- features of young adult Hispanic psychiatric
sonality and Social Psychology, 51, 1173–1182. in-patients: The so-called “Puerto Rican syn-
Beck, A. T. (1972). Measuring depression: The drome.” Military Medicine, 156, 351–354.
depression inventory. In T. A. Williams, M. Crittle, M. C. (1996). Stress, social support, and
M. Katz, & J. A. Shields (Eds.), Recent advances well-being in midlife Black and White
in the psychobiology of the depressive illnesses (pp. women. Dissertation Abstracts International B,
299–302). Washington, DC: U.S. Govern- 57(6-B), 4088.
ment Printing Office. Culbertson, F. M. (1997). Depression and gen-
Betancourt, H., & Lopez, S. R. (1993). The study der: An international review. American Psy-
of culture, ethnicity, and race in American chologist, 52, 25–31.
psychology. American Psychologist, 48, Derogatis, L. R., & Melisaratos, N. (1983). The
629–637. Brief Symptom Inventory: An introductory
Blazer, D. G., Kessler, R. C., McGonagle, K. A., & report. Psychological Medicine, 13, 595–605.
Swartz, M. S. (1994). The presence and dis- Dressler, W. W., & Badger, L. W. (1985). Epide-
tribution of major depression in a national miology of depressive symptoms in Black
community sample: The National Cormor- communities: A comparative analysis. Journal
bidity Survey. American Journal of Psychiatry, of Nervous and Mental Disorders, 173, 212–220.
151, 979–986.
Escobar, J. I., Rubio-Stipec, M., Canino, G., &
Brislin, R. (1993). Understanding culture’s influence Karno, M. (1989). Somatic Symptom Index
on behavior. San Diego, CA: Harcourt Brace (SSI): A new and abridged somatization con-
Jovanovich. struct: Prevalence and epidemiological corre-
Brown, C., Schulberg, H. C., & Madonia, M. J. lates in two large community samples. Journal
(1996). Clinical presentations of major de- of Nervous and Mental Disease, 177, 140–146.
pression by African Americans and Whites in Fabrega, H., Mezzich, J., & Ulrich, R. F. (1988).
primary medical care practice. Journal of Af- African American–White differences in psy-
fective Disorders, 41, 181–191. chopathology in an urban psychiatric popu-
Brown, D. R., & Gary, L. E. (1985). Predictors of lation. Comprehensive Psychiatry, 29, 285–297.
depressive symptoms among unemployed First, M. B., Spitzer, R. L., Gibbon, M., & Wil-
Black adults. Journal of Sociology and Social Wel- liams, J. B. W. (1997). Structured Clinical Inter-
fare, 12, 736–754. view for DSM–IV Axis I Disorders—Clinician Ver-
Brugha, T. S., Bebbington, P. E., MacCarthy, B., sion (SCID–CV). Washington, DC: American
Potter, J., Sturt, E., & Wykes, T. (1987). Social Psychiatric Press.
networks, social support and the type of de- Gant, L. M., Nagda, B. A., Brabson, H. H., Ja-
pressive illness. Acta Psychiatrica Scandinavica, yaratne, S., Chess, W. A., & Singh, A. (1993).
76, 664–673. Effects of social support and undermining on
Brugha, T. S., Bebbington, P. E., MacCarthy, B., African American worker’s perceptions of co-
Sturt, E., Wykes, T., & Potter, J. (1990). Gen- worker and supervisor relationships and psy-
der, social support and recovery from depres- chological well-being. Social Work, 38, 158–
sive disorders: A prospective clinical study. 164.
Psychological Medicine, 20, 147–156. Golding, J. M., Aneshensel, C. S., & Hough, R. L.
Chisholm, J. F. (1996). Mental health issues in (1991). Responses to depression scale items
African-American women. Annals of the New among Mexican-Americans and non-His-
York Academy of Sciences, 789, 161–179. panic Euro-Americans. Journal of Clinical Psy-
Clark, R., Anderson, N. B., Clark, V. R., & Wil- chiatry, 47, 161–171.
liams, D. R. (1999). Racism as a stressor for Gotlib, I. H., & Hammen, C. L. (1992). Psychologi-
African Americans: A biopsychosocial model. cal aspects of depression: Toward a cognitive–in-
American Psychologist, 54, 805–816. terpersonal integration. Chichester, UK: Wiley.
Cohen, S., Karmarck, T., & Mermelstein, R. Guernaccia, P. J., Angel, R., & Worobey, J. L.
(1983). A global measure of perceived stress. (1989). The factor structure of the CES-D inYou can also read
