Conceptual framework of the Spectrum Study, adapted from Cooper et al.18
Identification of depression by primary care physicians according to the level of depressive symptoms as measured by the Center for Epidemiological Studies Depression Scale (CES-D) and stratified by patient ethnicity. Data are from the Spectrum Study (2001-2003).19,20
Gallo JJ, Bogner HR, Morales KH, Ford DE. Patient Ethnicity and the Identification and Active Management of Depression in Late Life. Arch Intern Med. 2005;165(17):1962-1968. doi:10.1001/archinte.165.17.1962
Black Americans are more likely to obtain mental health care from a primary care physician than from a mental health specialist. We investigated the association of ethnicity with the identification and active management of depression among older patients.
Cross-sectional survey of 355 older adults with and without significant depressive symptoms. At the index visit, the physician’s ratings of depression and reports of active management were obtained on 341 of the 355 patients who completed in-home interviews.
Older black patients were less likely than older white patients to be identified as depressed (unadjusted odds ratio [OR], 0.40; 95% confidence interval [CI], 0.25-0.63) and their depression was less likely to be actively managed in the 6 months before the interview (unadjusted OR, 0.63; 95% CI, 0.19-2.16). In multivariate models that controlled for potentially influential characteristics such as patient age, sex, marital status, level of education, functional status, physical health, severity of depressive symptoms, severity of anxiety symptoms, attitudes about depression, number of office visits in the last 6 months, and the physician’s rating of how well they knew the patient, the associations of identification (OR, 0.25; 95% CI, 0.17-0.39) and management (OR, 0.57; 95% CI, 0.19-1.77) with patient ethnicity remained substantially unchanged.
Our study calls attention to the role ethnicity may play in the identification and active management of depression among older primary care patients.
The US Surgeon General renewed attention to the problem of unrecognized and untreated depression among older adults,1 stating that the primary care setting may provide the only opportunity in which serious depression and other mental conditions that affect elderly minority patients can be addressed.2 Primary health care is pivotal for the initial identification and management of depression among older patients.3 Black patients are more likely than white patients to obtain mental health care from a primary care provider. 4- 7 The increase in antidepressant use among older adults has occurred mostly among white patients.8 Despite the increase in antidepressant use among older adults, once in care, black patients appear to be less likely than white patients to receive appropriate diagnoses9 or to receive treatment thought to be effective for depression.5,10 Black patients have been reported to be less likely than white patients to adhere to a depression treatment regimen once it is initiated,11 even though no difference in treatment response according to ethnicity among patients who accept and sustain treatment has been identified.12 None of the cited studies focused on older primary care patients.
At the outset, we acknowledge that terms of race and ethnic origin are coarse markers of complex social and behavioral patterns. Ethnicity refers to a common heritage shared by a particular group.13 Culture is broadly defined as a common heritage or set of beliefs, norms, and values.2 We recognize that designations of ethnic status imply a homogeneity within groups that is a simplification. 14,15 Consistent with current publications of the National Institutes of Health,16 we use the terms black to include individuals of African, African American, and African Caribbean descent and white to include individuals of European descent. Any differences we observe across ethnic groups are likely to represent measured and unmeasured differences in social class, exposures, health beliefs and practices, and other characteristics.17
The goal of this investigation was to examine the characteristics of older patients identified by their primary care physicians as depressed, with a focus on the ethnicity of the patient. We hypothesized that black patients might have a different experience regarding depression identification and management than white patients in primary care practice. Our conceptual framework was a simplified version of the model suggested by Cooper and colleagues18 (Figure 1). In this model, attitudes about depression and familiarity with the practice mediate the relationship between patient ethnicity and the identification and active management of depression by primary care physicians. In the work of Cooper and coworkers,18 attitudes about depression and its treatment did not completely explain the relationship between patient ethnicity and acceptance of treatment for depression.
The overarching goal of the Spectrum Study was to characterize how depression presents in older primary care patients. The design of the Spectrum Study was a cross-sectional survey of patients and physicians. In all, 47 physicians (28 family physicians and 19 internists) from 13 practices contributed patients who participated in the Spectrum Study. 19,20 Practices were selected to achieve a substantial representation of black patients. Experienced agency interviewers worked with office staff at each practice to identify all patients 65 years and older. All such patients who came to a participating office were approached and asked to join in a study of health and aging. The study protocols were approved by the Institutional Review Board of the University of Pennsylvania School of Medicine, Philadelphia. A certificate of confidentiality was obtained from the Department of Health and Human Services as an additional safeguard.
At the index visit, the physician was asked to complete a brief assessment of the patient’s condition at that visit. The physician rated the patient’s level of depression at that visit on the following 4-point scale: none at all, mild, moderate, or severe. For this investigation, physician identification was defined as including ratings of mild, moderate, and severe depression. Next, the physician was asked to indicate whether any of the following were performed for the patient at the index visit or at any time in the past 6 months: “provided counseling/supportive listening for depression,” “tried to refer to a mental health specialist,” and “prescribed psychotropic medicine for depression, anxiety, or sleep.” Physicians were asked to provide a reason for not prescribing medication or counseling if the patient was thought to be depressed by checking as many reasons as applied from a list or by writing in a reason of their own. Physicians were asked to rate how well they knew the patient.
We obtained information from the respondents on age, sex, ethnicity, marital status, living arrangements, level of educational attainment, and the number of visits made to the practice for medical care within 6 months of the index visit. Patients who denied being Hispanic or Latino/Latina were asked to select from the following choices read to them: American Indian or Alaska Native, Asian, black or African American, native Hawaiian or other Pacific Islander, white, or other. Patients who responded that they belonged to another ethnic group were excluded from this analysis. Persons who self-identified as African American were classified as black for the purposes of this investigation.
The Center for Epidemiological Studies Depression Scale (CES-D) was developed for use in studies of depression in community samples. 21- 23 The standard CES-D questionnaire contains 20 items and has been used in studies of older adults. 24,25 In this study, we used the CES-D finding as a continuous score, but we also identified patients whose CES-D score was 16 or higher as depressed. The CES-D appears to be a valid measure of depressive symptoms among persons from differing ethnic groups. 26- 28 The Beck Anxiety Inventory was used as a continuous score to measure anxiety symptoms. 29- 31
Questions from the 36-Item Short-Form Health Survey (SF-36) were used to assess functional status. 32,33 The SF-36 has been used in studies of outcomes of patient care, 32- 36 and appears to be reliable and valid even in frail elders.37 The SF-36 was scored using previously described techniques.38 Baseline medical comorbidity was measured by summing the lifetime presence of 12 chronic diseases or conditions, including myocardial infarction, angina, congestive heart failure, high blood pressure, diabetes mellitus, osteoarthritis, stroke, cancer, Parkinson disease, hip fracture, and vision and hearing problems. We asked patients whether they agreed or disagreed with 3 statements about depression and its treatment.39 The statements were “I believe depression is a medical problem,” “If my doctor told me I had depression, I could accept that,” and “I would take a medicine for depression if my doctor told me to.”
We compared the characteristics of persons who self-identified as black with those of persons who self-identified as white, using 2-tailed t tests or χ2 tests for continuous or categorical data as appropriate. Our primary dependent variable was identification of the patient as depressed by the physician at the index visit. To adjust our estimates for potentially influential covariates and practice-clustering effects, we used generalized estimating equations for binary outcomes.40 We also examined identification in relation to whether or not the respondents scored 16 or higher on the CES-D,21 using the odds ratio (OR) as a measure of association. Consistent with our conceptual framework, we introduced terms in the multivariate models to represent age, sex, marital status, level of educational attainment, functional status, physical health, and depressive and anxiety symptoms. Subsequent models included terms for attitudes about depression and its treatment and familiarity with the practice (as represented by the patient’s report of the number of visits made to that practice in the 6 months before the interview and the physician’s rating of how well they know the patient) that were hypothesized to be mediators in our conceptual framework. In subsequent analyses, we restricted our attention to the persons whom the physician identified as depressed, with report of active management as the dependent variable. Finally, we tabulated physician reasons for not prescribing medication or counseling for patients identified as depressed.
Table 1 compares the characteristics of black and white patients in the sample. Proportionately more black patients were women and unmarried. Three persons self-identified as “other” for ethnicity and were excluded. Black patients were significantly more likely than white patients to report poorer physical and social functioning. Proportionately fewer black patients were identified as depressed by physicians or were reported to have depression actively managed within 6 months of the index visit.
Black patients who were identified as depressed, when compared with white patients identified as depressed, were more likely to report worse physical and social functioning and bodily pain (Table 2). Physicians indicated that they knew depressed black patients equally as well as white patients, but physicians were more likely to report that they knew black patients well who were identified as depressed than they did black patients who were not identified as depressed. Black patients who were not identified as depressed by the physician had poorer physical and role emotional functioning compared with white patients not identified as depressed. Figure 2 illustrates the detection rates according to CES-D strata and ethnicity. Black patients were less likely to be identified as depressed across the entire range of CES-D strata. Among persons who scored above the threshold on the CES-D, black patients were less likely than white patients to have been identified as depressed (OR, 0.48; 95% confidence interval [CI], 0.26-0.88).
Black patients whose depression was actively managed reported statistically significantly higher depression scores than white patients whose depression was actively managed, and poorer functioning on several measures (Table 3). Black patients whose depression was actively managed made more visits to the primary care office than white patients whose depression was actively managed.
After adjustment with generalized estimating equations to account for clustering and potentially influential covariates, including severity of depression symptoms, black patients were 0.25 times as likely (95% CI, 0.17-0.39) to have been identified as depressed (Table 4). Black patients were about half as likely as white patients to have received active management of depression (OR, 0.57), adjusting for potentially influential covariates, although confidence bounds of the point estimate included the null (95% CI, 0.19-1.77).
Physicians provided reasons for not actively managing the patient’s depression in the 6 months before the index visit among the patients they identified as depressed. The most common reason was that the depression was too mild (6 of 12 black patients and 9 of 25 white patients). The next most commonly cited reasons were that the depression represented a temporary adjustment reaction (2 of 12 black patients and 4 of 25 white patients) and that the patient would be reluctant to accept the diagnosis of depression (3 of 12 black patients and 2 of 25 white patients).
In this primary care sample, black patients were less likely than white patients to have been identified as depressed. This association persisted even after controlling for potentially influential variables, including severity of depressive symptoms and level of functioning. Black patients were also less likely than white patients to have depression that was being actively managed after controlling for potentially influential variables; however, the association approximated but did not reach standard levels of statistical significance. Black patients identified by their physician as depressed whose depression was being actively managed were more likely than white patients to be functionally impaired. These results are all the more remarkable because we have asked physicians to report on their care of persons whom they identify as depressed. Our study builds on previous work 9,18,41 and suggests that patient ethnicity may play a role in the identification and active management of depression among older adults who present in the primary care setting.
Before putting our results in clinical and research context, the limitations should be discussed. Our study was based on practices that participated in a research study and so may not be representative of all practices. The responses to the survey instruments do not necessarily reflect the actual discussion the patient had with the physician during the office visit when the patient was rated by the physician. In other words, we cannot be certain of the extent to which patients of different ethnic groups may express their symptoms differently during physician encounters. We did not have ratings of patient behavior or symptoms from clinical observers other than the patient’s primary care clinician. Despite these limitations, our study design is associated with several strengths, and the findings deserve attention. Although limited to practices in a single geographic area (Baltimore, Md), we attempted to create a representative sample by random calling of physicians based on a sampling frame provided by a professional organization, we included multiple practices, and we sampled from urban and suburban areas. The data from physicians are nearly complete (97% of the participants were rated by their physician). We were able to adjust our measures of association for ratings of how well the physicians knew the patient and other potentially influential covariates. Because we linked patient data to whether the physician reported that they had actively managed the patient’s depression within 6 months of the interview, we did not have to depend on medical chart reviews.
The Institute of Medicine report “Unequal Treatment”42 reviewed the medical literature on the extent and sources of disparities in health behaviors and outcomes across ethnic groups. Despite the observed variation in recognition and treatment of psychiatric disturbances according to ethnicity,9 empirical studies have not typically focused on the patient-physician encounter in studying observed differences in rates of antidepressant use, seeking specialty mental health treatment, or discussing a mental health problem with a general medical health care provider. 1,2,42
Few studies have focused on older adults to assess the relationship between physician ratings, patient characteristics, and identification and active management of depression. Miranda and Cooper43 found that Latino and African American patients were no less likely to be offered treatment for depression by primary care physicians, yet they were less likely to actually receive appropriate care for depression. Few older adults were included in that study (eg, only 2 African American patients were 65 years or older). Borowsky and colleagues44 reported that black patients were less likely to be identified by primary care physicians as having a mental health problem. That study was not focused on older adults or on the specific diagnosis of depression. Sirey and coworkers45 reported that white patients were much more likely than black patients to receive a recommendation for an antidepressant, but their study did not include adults older than 65 years and did not focus on the primary care setting. Studies of older adults in public housing46 and in long-term care47 showed that older African Americans were less likely to use available community mental health services, but no study of recognition has focused on older adults in primary care settings as we have done here.
For clinicians, it may be easier to determine whether a patient has a symptom of depression than to determine the severity of the symptom. When judgment about uncertain elements of the clinical examination is required, patient characteristics can have a greater effect on the final assessments. Physicians caring for black patients should be aware that there is a tendency for depressive symptoms to be judged as not severe. Physicians might introduce this fact into their interviews when appropriate and determine whether this leads to a different conversation with their black patients. Acknowledging that the possibility of a discrepancy occurs may be the first step toward improving the management of depression.
The recognition and management of depression in primary care settings is a negotiated process between the patient and the primary care physician. Given the competing demands of practice,48 the common uncertainty inherent in the diagnosis of depression in primary care settings, and differing perceptions of the physician-patient encounter across ethnic groups, 41,49 physicians may take their cue for identification and management decisions from their assessment of the likelihood that a patient will accept the diagnosis or treatment recommendations. In addition, physicians may interpret symptoms of persons as too mild to treat when the patient is black. Physicians and patients may collude not to treat depression, ascribing symptoms to other causes such as social conditions. We are performing analyses of open-ended interviews of the sample to understand what older adults think about the nature, treatment, and outcome of depression. Physicians and patients in primary care settings come to the encounter with attitudes, expectations, and values that may differ markedly from physician to physician and between physician and patient. Sorting out the multiple sources of disparity will require fine-grained studies of how characteristics of providers, patients, and health care systems influence the interplay of physician, patient, and environment.
Correspondence: Joseph J. Gallo, MD, MPH, Department of Family Practice and Community Medicine, University of Pennsylvania, 3400 Spruce St, 2 Gates Pavilion, Philadelphia, PA 19104 (email@example.com).
Accepted for Publication: May 23, 2005.
Financial Disclosure: None.
Funding/Support: This study was supported by grants MH62210, MH62210, and MH67077 from the National Institute of Mental Health (NIMH), Rockville, Md (the Spectrum Study); Mentored Patient-Oriented Research Career Development Award MH67671-01 from the NIMH (Dr Bogner); and funding for a Robert Wood Johnson Foundation Generalist Physician Scholar (2004-2008) (Dr Bogner).