Effects of Primary Care Depression Treatment on Minority Patients'Clinical Status and Employment FREE

Ethnic minorities compose an increasingly large portion of the US population,roughly 30% in 2000, with projections nearing 40% by 2050.¹ Meetingthe medical needs of this growing population is an important priority forour health care system, yet often data on outcomes for ethnic minority populationsare unavailable. Because cultural differences may be particularly importantto mental health, determining outcomes of mental health care for these largeminority populations is crucial.

Little is known about outcomes of mental health interventions for ethnicminorities. The Surgeon General's Report on Mental Health² describesa range of well-established, efficacious treatments for most mental disorders,but ethnic minorities have largely been missing from the scientific inquirythat defines appropriate mental health care. For example, the efficacy studiesused by the American Psychiatric Association and the American PsychologicalAssociation to establish guidelines for care for depression included only27 identified African American, 2 Asian/Pacific Islanders, and no Latino orNative American/Native Alaskans.^3(p35) Only a few studies of depression care for minorities have been published,but none are controlled and the sample sizes are far too small to establishthe response of minorities to care.³

Some evidence suggests ethnic minorities may respond differently tomental health interventions.⁴ For example,Sue et al⁵ analyzed changes in Global AssessmentScale scores in patients in the Los Angeles County Mental Health System andfound that African Americans had lower positive treatment outcomes than didall other ethnic groups. Brown et al⁶ foundequal clinical but poorer functional outcomes for treated depressed AfricanAmericans compared with white participants. Very little data are availableto explain such differences in outcomes. Historically, race was considereda biological category based on common inborn biological traits such as skincolor and eye shape; these biological differences were assumed to underlieracial differences. However, there are no biological underpinnings for dividingpeople into distinct racial categories.^7- 8 Noconsistent racial groupings emerge when people are sorted by physical andbiological characteristics. Despite this finding, a small body of literaturesuggests that environmental and genetic differences may result in AfricanAmericans, Asians, and Latinos metabolizing depression medications, on average,differently from those who are not minorities.⁹

Apart from biological differences, these groups might differ as a resultof shared history. Because of unique histories (ie, slavery, legal statusin the United States), minorities tend to be poorer, on average, than thosewho are not minorities,¹ experience immigrantbackgrounds more frequently than do nonminorities (Latinos, Asians), and aremore likely to perceive discrimination in their lives.¹⁰ Culturaldifferences potentially could make psychotherapeutic interventions differentlyeffective for minorities. For example, if Latino culture encourages low levelsof assertiveness, outcomes of psychotherapy aimed at increasing assertivenessmay differ for those from Latino culture compared with those from culturesencouraging assertiveness. In this study, we attempt to identify whether depressioninterventions have universal applicability or work well only with the majoritypopulation.⁴

Understanding outcomes of care for ethnic minorities requires examiningthe context for this care. Minorities are particularly unlikely to receivecare in psychiatric settings in which most efficacy trials occur. For example,less than 16% of the African Americans diagnosed as having a mood disordersee a mental health specialist³ and less than9% percent of the Mexican Americans with a mental disorder seek care frommental health specialists.¹¹ Minorities whohave depression are more likely to be seen in primary health care settingsthan in psychiatric settings.^3,9 Documentingthe effect of care for depression among minorities would provide a scientificbasis for improving rates of detection and treatment.

Because minority populations generally have not participated in efficacytrials, data on outcomes of usual care practice could provide evidence onthe effectiveness of depression care for this population. However, observationalstudies are subject to strong selection bias, such that patients with moresevere illnesses are much more likely to get care than are those with lesssevere illness. Furthermore, sources of selection bias may differ betweenminority and nonminority populations. This study examines naturalistic outcomesof care for minority and nonminority populations but relies on instrumentalvariables analysis, a technique that allows causal inference when selectionbias is present.^12- 16 Specifically,the instrumental variables approach uses a popular technique in economic studiesthat controls for bias in selection without requiring a knowledge of the sourceof bias. Regression approaches can control for known biases through covariates,but not all sources of bias are identifiable in naturalistic studies. However,instrumental variables analyses can account for unknown biases when a variablecan be identified that is related to outcomes only through its effect on treatment.This variable must not have an independent effect on outcomes. By using suchan instrument in this study, we determine the effect of appropriate care forminority and nonminority participants, removing the bias associated with selection.

We use data from a randomized trial of quality improvement (QI) interventionsfor depression in which naturalistic practice conditions were preserved, thatis, usual providers, patients, payment mechanisms, and patient and providerchoice of treatment. The QI interventions were conducted by a cross-nationalsample of managed care organizations and were designed to improve the qualityof care they provided for depression within their organization. The participatinghealth care practices, using materials developed by our research team, educatedtheir providers about depression care, used physician extenders (depressionnurse specialists) to educate and guide patients into depression care, andworked in teams to improve the quality of their depression care. These QIinterventions improved quality of care as well as clinical, quality of life,and employment outcomes for depressed patients over 12 months.¹⁷ Furthermore,increases in quality of care were similar across ethnic groups; however, differencesemerged in the effect of QI on outcomes.¹⁸ Ethnicminorities had stronger clinical responses to QI; whereas, employment improvedmore for nonminority participants. These earlier results examine the differentialeffect of the overall QI interventions but do not provide an estimate of differentialeffect of appropriate depression treatments. In this article, we focus onthe outcomes of appropriate depression care (which many patients did not receiveeven under the interventions) as reported at a 6-month follow-up assessmentand provide data separately for Latino, African American, and nonminoritypatients whenever possible.

Partners in Care, a Patient Outcomes Research Team study sponsored byAgency for Health Research and Quality, is a group-level, randomized, controlledtrial of practice-initiated QI programs for depression.¹⁵

Six managed care organizations, selected to be diverse in geographyand organization, as well as to oversample Mexican Americans, participatedin this study. Grouped practice clusters formed matched blocks for randomization.The groups were matched on patient demographics, clinical specialty, and distanceto mental health providers. Nine blocks, 1 for each of 6 regions and 3 forthe seventh (stratified by low, intermediate, or high percentage of MexicanAmerican patients) were created. The clinics clusters were randomized by blocksto usual care or 1 of 2 QI conditions.

Primary care clinicians were recruited before learning the assignmenttheir clinic would receive; 97% (n = 181) agreed to participate. The clinicianswere 87% internists or family practice physicians and 13% nurse practitioners;32% were members of a minority group, including 2% African American, 18% Latino,and 12% other minorities.

Study staff screened 27 332 consecutive patients over 5 to 7 monthsbetween June 1996 and March 1997. Patients were eligible if they intendedto use the clinic as a source of care for the next year, were older than 17years, did not have an acute medical emergency, spoke English or Spanish,and had either insurance or a public-pay arrangement that covered the interventioncare. Eligible patients were screened for depression using the "stem" itemsfor major depressive and dysthymic disorder from the 12-month Composite International Diagnostic Interview, CoreVersion 2.1,¹⁹ and items assessing depressedsymptoms in the past month. Patients learned of their intervention statusfollowing enrollment.

Of those completing the screening, 3918 were potentially eligible. Ofthese, 2417 confirmed insurance eligibility; 241 were found ineligible. Ofthose who read the informed consent, 1356 (79%) enrolled: 443 in usual care,424 in QI medications (QI-meds), and 489 in QI therapy (QI-therapy). Herein,we examine 938 participants with probable major depression or symptoms ofdepression with a history of major depression. We do not examine those withsymptoms only so that we focus on a sample in need of depression treatment.Ethnic breakdown of participants was: 601 white, 257 Latinos, 56 African American,and 24 Asian or Native American. The enrolled patients completed the Composite International Diagnostic Interview to determinedepression diagnoses and the presence of comorbid anxiety and a telephoneinterview to determine income, wealth, and employment. Self-administered mailsurveys were obtained at baseline and every 6 months for 2 years. Survey responserates were 95% for the telephone interview, 88% for the baseline survey, and83% for 6- and 12-month surveys.

All QI materials are available from RAND. (Available at: www.rand.org/organization/health/pic.products/order.html.) The implementation involved resources from the practices. The practicescommitted to involving existing staff in implementing the QI procedures. Inaddition, the study provided a payment of up to half of the estimated practiceparticipation costs ($35 000-$70 000). The investigative team providedpractices with training and resources to initiate and monitor the QI programsaccording to local practice goals and resources. In essence, the practiceswere trained to help themselves improve care under naturalistic practice conditions.Patients and clinicians retained choice of treatment, and their use of interventionresources was optional. The study served as an external disease managementfirm, designing the materials, hosting initial training, and offering limitedsupport during implementation.

The QI interventions were designed to increase access and adherencewith some form of appropriate care for depression, either medication or psychotherapy.Care as usual involved mailing practice guidelines for depression to the clinics.

For both QI-meds and QI-therapy, local practice teams were trained ina 2-day workshop to provide clinician education through lectures, academicdetailing, or audit and feedback, and to supervise the QI interventions, aswell as conduct QI team oversight. Designated practice nurses were trainedas depression specialists, including assisting in initial patient assessment,education, and motivation for treatment. Practices were provided with patienteducation pamphlets and videotapes, patient tracking forms, and clinicianmanuals and pocket reminder cards and were encouraged to distribute them.The materials described guideline concordant care for depression and presentedpsychotherapy and antidepressant medication as equally effective.

In the QI-meds group, trained nurses were available to provide follow-upassessments and support for adherence to care for 6 months (with half thepatients randomized to 12 months). In the QI-therapy group, the study trainedlocal psychotherapists to provide individual and group cognitive behavioraltherapy for 8 to 12 sessions and provided them with patient and therapistmanuals.^20- 21 The organizationsreduced co-pay for psychotherapy to the primary care amount (ie, $0-$10 vs$20-$30). In usual care, no study resources were available. Patients who receivedQI-therapy could also receive antidepressants but did not have access to theQI-meds group's study resources. Patients who received QI-meds did not haveaccess to the study-sponsored cognitive behavioral therapy therapists butcould use other local therapists.

Within the overall aim of improving rates of appropriate care, the interventionshad a secondary aim of ensuring inclusion of minorities in opportunities forcare. Accordingly, experts in mental health interventions for minorities participatedin designing the QI educational materials. All QI intervention materials wereavailable in English and Spanish; Latino and African American providers wereincluded in videotaped materials for patients. Information regarding culturalbeliefs and ways of overcoming barriers to care for Latino and African Americanpatients were included in provider training materials (see "Culture Trainingfor Providers" box). Furthermore, materials provided to improve psychotherapyfor depression were developed for ethnically diverse (English-language materials¹⁸) and Latino (Spanish-language materials¹⁷)patients.

Appropriate treatment in the first 6-months of follow-up was measuredby survey items that assessed whether the respondent had 4 specialty counselingvisits or more or used antidepressants for at least 2 months at or above theminimum dosage recommended in the Agency for Healthcare Policy Research practiceguidelines,²² adapted to include newer antidepressants.We note that most patients either had appropriate care or almost no mentalhealth care. In particular, 90% of those using appropriate medication for2 months used it for 6 months; similarly, the mean (SD) number of visits amongpatients with at least 4 therapy visits was 12.5 (0.8).

Health outcomes at 6 months include probable depressive disorder, basedon a repeat of the screener items, and the global mental health scale of theMedical Outcomes Study 12-item Short Form.²³ Inaddition, we analyzed patients' self-reported employment status at 6 months.

Covariates collected at baseline include age, sex, marital status, educationalattainment, household wealth, employment status, medical comorbidity, depressivedisorders status, health-related quality of life, practice site, and the presenceof comorbid anxiety disorder (Table 1).We also controlled for the number of days between enrollment and the follow-upsurvey.

We used univariate and bivariate analyses to describe the sample andcompare patients who did or did not receive appropriate care in the first6 months. We used multivariate regression models to examine the effects ofappropriate care on outcomes. We expected that unobserved differences by treatmentstatus would remain even after controlling for baseline patient characteristics.We used the method of instrumental variables to account for these differences.^10- 14 Thismethod relies on identification of an instrument that predicts the probabilityof treatment but that has no independent effect on outcomes except throughtreatment; this approach helps to separate the effect on outcomes owing totreatment from the effect owing to unobserved characteristics.

We used randomized intervention status (case vs control) as the instrument,which meets the conditions for valid instruments specified by¹⁰:(1) random assignment, (2) the instrument affects treatment rates ("nonzeroaverage casual effects"), (3) the instrument is unlikely to affect outcomesexcept through treatment ("exclusion restriction"), and (4) patients withpositive outcomes in usual care would likely have positive outcomes underthe intervention ("monotonicity"). Each assumption seems plausible here.

We estimate a multivariate regression with each outcome as the dependentvariable, controlling for whether patients received appropriate care and thecovariates mentioned earlier; and we estimated a second regression with appropriatecare as the dependent variable, controlling for intervention status and thecovariates. For dichotomous outcomes (probable disorder, employment), theseequations were estimated simultaneously using a bivariate probit specification.²⁴ For the continuous global mental health scale, theseequations were estimated simultanously using the 2-stage least squares method.

We report outcomes adjusted for all covariates. We used the parameterestimates from the regression and each individual's actual value for covariatesto generate predicted values under the scenario that the patient receivedappropriate care and then under the scenario without appropriate care.^21,25 We then averaged the predictionsacross individuals under each scenario.

Although patients were clustered within providers and clinics, intraclasscorrelations are close to zero. For this reason, and because we know of nostandard methods to account for clustering in instrumental variables models,we report results unadjusted for cluster effects. We weighted these data forprobability of enrollment, attrition, and wave response, using standard logisticregression.²⁶ The nonresponse weights werethe inverse of the predicted probability of response from logistic regressionanalysis. Thus, the final sample was weighted to the population eligible totake the screener. We used multiple imputation for missing data at the itemlevel.^27- 28 We used 2-tailed testswith α = .05. Although we have 2 clinical outcomes, findings are ina consistent direction, so formal correction for multiple comparisons is tooconservative.

Initial results are given in Table1. As reported previously, patients in usual care practices wereless likely than those in QI intervention practices to receive appropriatecare during the first 6 months after the intervention.¹⁵ Minoritieswere receiving lower rates of appropriate care prior to the introduction ofQI interventions, and the QI interventions significantly improved the ratesof those receiving appropriate care by 9 to 20 percentage points for all patients(minorities and nonminorities) at 6 months.¹⁶ Ascan be seen in Table 1, approximately32.8% of the minority patients and 51.4% of white patients were getting appropriatecare by 6 months.

Patients getting appropriate care at 6 months differed from those notgetting appropriate care (Table 1).Minorities getting appropriate care were older, had attended college, andwere more likely to be unemployed than were those not receiving appropriatecare. Nonminorities receiving appropriate care were more likely to be femalethan those not getting appropriate care. Similarly, predicted clinical characteristicsdiffered among those receiving and not receiving appropriate care. Minoritiesreceiving appropriate care were more likely to have an anxiety disorder, poorerquality of life attributable to physical illness, and 3 or more chronic illnessesthan were those not receiving appropriate care. Nonminority patients receivingappropriate care were more likely to have an anxiety disorder, presence ofboth chronic and acute depression, a lifetime diagnosis of major depression,and a poorer perceived quality of life related to psychological problems comparedwith those not receiving appropriate care. These analyses suggest that differentfactors influence minority and nonminority patients' access to appropriatecare.

Table 1 also lists a simplebivariate comparison of outcomes at 6 months by treatment status. For minorities,those receiving appropriate care were more likely to be unemployed, but otheroutcomes did not vary by treatment status. For nonminorities, those receivingappropriate care reported a poorer quality of life. These findings indicatethat there was substantial selection into care; those patients who were fairingworse at baseline were more likely to receive appropriate care.

Table 2 summarizes the resultsafter accounting for patient covariates and selection into treatment throughthe instrumental variables method. According to these findings, approximately20% of minorities receiving appropriate care were likely to meet criteriafor probable depressive disorder at 6 months compared with nearly 70% of thosenot receiving appropriate care. The percentages were similar for the nonminorities.Neither global mental health functioning nor employment was significantlychanged as a result of appropriate care, although there was a trend (P = .06) for nonminorities to be employed more if receivingappropriate care (71% vs 52%).

Because the combined ethnic minority group contains diverse Latinos,African Americans, and others, we attempted to determine whether the resultswould hold for specific minority groups, recognizing the limitations imposedby our sample size. Table 3 replicates Table 1, separately for Latino, AfricanAmerican, and other minority patients, respectively. For the instrumentalvariables models examining the influence of appropriate care on probable disorder,we were able to replicate the analyses using the combined Latino and AfricanAmerican samples and excluding the other ethnic groups. Results were substantivelysimilar to those reported for minority patients in Table 2. For the continuous variable, global mental health qualityof life, we were able to replicate the analyses for Latino patients and forAfrican American patients, respectively. Results were also similar to thosereported for minority patients in Table2. The estimated mean score was 36.8 for Latinos with appropriatecare, compared with 46.4 for Latinos without, while the corresponding resultswere 38.6 vs 48.0 for African American patients. Because of limited samplesizes, these estimates may be unreliable; however, the similarity in the findingsfor Latinos and African Americans may suggest these groups respond similarlyto depression care.

In this article, we are able to examine the effect of depression careon ethnic minorities who have been poorly represented in efficacy studies.Using an instrumental variables approach, we control for selection bias todetermine the relative effectiveness of such care in common clinical settings.This study finds that appropriate care reduces the likelihood of probablemajor depression from approximately 70% to 20% for minority and nonminoritypatients. This suggests that care for depression in naturalistic settingswill dramatically reduce depression for multiethnic patients.

The effect of appropriate care for depression on employment status maydiffer for minorities compared with nonminorities. Appropriate care seemsto increase the likelihood of being employed for nonminority participantsbut not for minorities. This may be because appropriate depression care improvesthe functioning of minorities more than that of nonminorities. On the otherhand, both may improve equally, but minority individuals may be less likelyto capitalize on improvements in functioning by obtaining jobs because ofeducational and occupational preparedness differences. Further study is neededto understand these work findings and their implications for the effect ofdepression care on functioning of minorities.

The results of this article suggest that despite differences among minorityand nonminority patients, including potential social-historical-metabolicdifferences, depression interventions appear effective across groups. Unlikeearlier studies focusing on single-item outcomes judged by clinicians,⁵ this study with independent assessment of disorderstatus finds that minority individuals respond well to care. Unfortunately,minority individuals are much less likely to be receiving appropriate carethan are nonminority patients.³ These findingssuggest that cultural differences may play a major role in access to care,but a much less prominent role in response to care. Efforts to educate medicalstaff about talking with ethnic minority patients regarding depression careand encouraging them to engage in care will be important in reducing disparities.

Our findings also show differences in prediction of care by ethnic group.For nonminority participants, females and those with more severe depressionare most likely to get care. For minority populations, those who are older,more educated, unemployed, and have worse health are more likely to get care.These differences may reflect cultural values, such that minority individualswho are physically ill feel more comfortable accepting depression care thanthose who are less ill. On the other hand, employed minorities may have moredifficulty obtaining time off work for care than do others. Further understandingof factors influencing access to care for minority populations is needed.

Lack of information on the mental health needs and responsiveness ofethnic minority populations is a critical disparity. Too often, the best availableresearch on ethnic minorities consists of small studies that cannot be generalizedto today's increasingly diverse communities. This article provides evidencethat standard depression care is effective for improving depression amongdiverse ethnic minorities when provided in Spanish and with some sensitivityto the needs of minority populations. The accommodations for minorities inthis study included minimal cultural training (see "Culture Training for Providers"box) for providers as well as attention to including minorities in videotapedmaterials. Importantly, all materials were available in Spanish. This minimalattention to cultural factors led to substantial increases in care for ethnicminorities and our subsequent demonstration that clinical response to careis similar to that for nonminority patients.

Several limitations should be noted in this study. First, a diversegroup of ethnic minorities are considered together and may mask differencesin response to appropriate care by subpopulations. Analyses do confirm thatthe major findings seem to hold for Latinos and African Americans, but thesample of other minority populations is too small to assess. Second, technicallythe instrumental variables analyses method identifies the effects of treatmentfor patients who are likely to receive care under the QI intervention butnot under usual care; effectiveness may be different for patients who arelikely to be treated under practice conditions. Third, we were unable to adjustfor clustering of patients within clinics in these analyses because of limitationsin the statistical procedures. Fourth, although we studied a range of managedcare practices, different findings could apply in other settings. Fifth, appropriatecare as defined in this study is slightly below full guideline recommendations.

This study provides evidence that guideline-concordant care for depressionis clinically effective for ethnic minority patients, even when that careis provided under community-based practice conditions. Our findings provideuseful information for policy makers, as well as practices and clinicians,because they support the value of providing conventional depression treatmentsto minority populations. The findings suggest that efforts to develop "ethnic-specific"treatments for depression may be unnecessary, at least for Latinos and AfricanAmericans. However, that does not mean that treatment-providing strategies,such as QI interventions, do not need to be modified for minority populations.Indeed, our Partners in Care study modified our QI intervention approach tofacilitate higher rates of use of standard treatments by minority populations.Significant need exists to encourage primary health care providers to detectand treat depression in minority patients. This may include overcomingserious barriers to care, such as inability of minority workers to take timeoff of work to come to treatment. Our results suggest that using effectivetreatments for depression in minority patients will result in a similar levelof improvement as for nonminority populations. By increasing rates of identificationand treatment for depression among ethnic minorities, disparities in carethat are due to low rates of treatment could be reduced or eliminated.

Several aspects of Latino culture affect the willingness of Latino patientsto engage in treatment. By attending to these cultural aspects, you will findthat Latinos are more likely to agree to treatment for depression. The firstcultural factor to consider is that of respecto, thatis, the expectation that males and older Latinos will be treated with respect,particularly by younger persons. This cultural characteristic suggests thatit is particularly important to use titles of respect when speaking with Latinosin general, but with older Latinos in particular. Using the formal usted to connote respect is particularly useful if a younger intervieweris working with a patient.

Latinos are also generally taught to respect powerful others, such asphysicians and nurses. As a result, it may be necessary to work particularlyhard with Latino patients to make sure that they have indicated any concernsthat they may have about getting treatment for depression. They are more likelythan nonminority patients to say "Yes, Doctor" without clearly stating reservationsunless they are given clear permission by the physician or nurse to stateany concerns. If these concerns are not elicited, the patient has a much higherlikelihood for failing to show up for treatment

Latinos, particularly those who are less acculturated, may be influencedby the cultural concepts of machismo and marianismo. Machismo refers to the general dominance of the male inthe family, with both the responsibility for providing for and protectingothers in the family. When dealing with more traditional Latino families,it may be necessary to obtain the consent of the male head of the householdfor a Latina to obtain treatment. If this is the case, bringing the husband-fatherin to discuss treatment options and planning may facilitate getting the wifeor daughter in for treatment. The cultural concept of marianismo refers to the belief that suffering for women is directly relatedto spirituality, such that the "long-suffering" woman may be revered. Becauseof this concept, traditional Latinas may need extra encouragement that theyshould not be suffering and do not deserve to feel so bad. Often they arenot as motivated to help themselves as they might be to help others. Therefore,reminding a traditional Latina that she needs to take care of herself to takecare of others may be particularly useful.

Latinos may be used to "warmer" interactions with others than are non-Latinos.In this regard, the concept of simpatia is particularlyimportant. That is, traditional Latinos tend to like warm relationships, whereothers are showing clear concern for them and their family members. To besthelp Latinos enter treatment, several important steps should be undertaken.An effort to remember the names of important family members of the patientscan be helpful. Remembering to ask about the patient and others in his orher family is helpful. Making sure that you ask the patient if he or she iscomfortable or needs anything can be useful, as can providing a beverage duringa long interview.

Finally, Latinos may have more barriers to overcome to come to treatment.For example, many Latinas may not have access to babysitting to come to treatment,or others may be unable to ask for time away from work to come to treatment.Working with patients to help them overcome these barriers is an essentialcomponent to offering culturally sensitive treatment.

The key to understanding how to evaluate African Americans is to understandthat they are a diverse and heterogeneous group. Stereotypes are inaccurateand harmful. There are vast socioeconomic, regional, and subgroup ethnic identitydifferences (Creole, Cajun, Caribbean, and others). Although it is unnecessaryfor a provider to have knowledge of specific differences among these groups,it is important that the provider make an effort to understand his or herpatient's particular cultural beliefs about illness and treatment.

Because of the unique history of African Americans and long-standingstruggles with prejudice and discrimination, it is also important for providersto realize that many come to the medical encounter with fears about mistreatmentand/or poor treatment. These fears have been exacerbated by recent media reportsof substandard care for African American patients with renal and cardiac disease.

Surveys of African American patients reveal a preference for providerswho are open, engaging, and direct in style. They are less likely to trustclinicians who appear to be distant or indifferent. They also prefer providerswho are careful to elicit personal values and preferences for both diagnosisand treatment.

Studies of help-seeking behavior have shown that many African Americansare more likely to seek help from spiritual advisors (ministers or religiouscounselors) for emotional problems. Queries about emotional health should,therefore, include questions about history and experience with spiritual counseling.Because a higher percentage of African American patients have low socioeconomicstatus, it is also essential that providers remember to elicit informationabout financial barriers to recommend care.

Correspondence: Jeanne Miranda, PhD, Health Services Research Center,University of California, Los Angeles–Neuropsychiatric Institute, 10920Wilshire Blvd, Suite 300, Los Angeles, CA 90024 (mirandaj@ucla.edu).

Submitted for publication October 25, 2002; final revision receivedMarch 31, 2003; accepted March 31, 2003.

This study was funded by the John D. and Catherine T. MacArthur Foundation,Chicago, Ill; grants R01-HS08349 from the Agency for Healthcare Research andQuality, Washington, DC, and 5R01MH57992 from the National Institute of MentalHealth, National Institutes of Health, Bethesda, Md. The writing of this articlewas supported by grants AG-02-004 from the Resource Centers for Minority andAging Research, National Institute of Aging, National Institutes of Health,Bethesda; 5P01-HS1-0858:03 from the US Department of Health and Human Services,Washington; MH59876 from the Latino Research Program Project, National Instituteof Mental Health, Bethesda; and MH546230-06 from the Research Center on ManagedCare for Psychiatric Disorders, National Institute of Mental Health, Bethesda(all grants given to Dr Miranda).