Patient Ethnicity and the Identification of Anxiety in Elderly Primary Care Patients

Abstract

OBJECTIVES

Our objective was to examine the role of ethnicity and primary care physician identification of anxiety among older adults.

DESIGN

A cross-sectional survey conducted between 2001–2003.

SETTING

Primary care offices in the Baltimore, Maryland area.

PARTICIPANTS

A sample of 330 adults aged 65 and older from Maryland primary care practices with complete information on psychological status and physician assessments.

MEASUREMENTS

Primary care physicians were asked to rate anxiety on a Likert scale. Patient interviews included measures of psychological status and patient use of psychotropic medications.

RESULTS

Older black patients were less likely than older white patients to be identified as anxious (unadjusted Odds Ratio (OR) = 0.34; 95% Confidence Interval (CI) [0.18, 0.64]) and were less likely to be taking psychotropic medications (unadjusted OR = 0.40; 95% CI [0.20, 0.81]). In multivariate models that controlled for potentially influential characteristics including depression and anxiety symptoms, the association of identification (OR = 0.30; 95% CI [0.15, 0.61]) with patient ethnicity remained significantly unchanged.

CONCLUSION

Older Black Americans were less likely to be identified as anxious compared to white patients by their primary care physicians. An understanding of the role of ethnicity in the identification of anxiety is important for the screening and management of anxiety in the elderly (207 words).

INTRODUCTION

The primary health care setting is pivotal for the identification and treatment of mental disorders, as the elderly remain underserved by mental health providers. In addition, the primary care setting may be even more important for the mental health of elderly minority patients, as black patients are more likely to seek mental health care from a primary care physician than a mental health specialist.^{1, 2} Much of the research examining psychiatric illness identification and management in black patients has focused on depression. Older black patients are less likely than older white patients to be identified as depressed or actively managed for their depression.³ The increase in antidepressant use among older patients has occurred mostly in white patients.⁴ In addition, black patients have been found to be less adherent to depression treatment regimens, even though no ethnicity-related treatment response difference has been identified.⁵

In the 2002 Institute of Medicine (IOM) Report commissioned by the United States Congress, Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care, a comprehensive review of the medical literature demonstrated significant health disparities in the treatment of illness and delivery of care across ethnic groups, with mental health being no exception.⁶ Despite this documented variability in the identification and treatment of mental illness according to ethnicity, there is a paucity of studies focusing on the doctor-patient encounter to explain the observed differences in the rates of presentation to a primary care provider, identification of mental illness, psychotropic medication use, or referral seeking for specialty mental health treatment.³ An analysis of 508 audiotaped outpatient doctor-patient encounters by Sleath and colleagues showed that physicians were more likely to minimize emotional symptoms of African Americans than of whites, which led to lower prescribing of antidepressant medication among African Americans.⁷ Cooper-Patrick et al., in a study of 1816 adult primary care patients, found that African American patients rated their office visits as less participatory than white patients.⁸

While the literature on the identification and management of depression in the elderly has been developing at a steady pace, the research on anxiety has lagged behind, despite its public health significance. Subthreshold late-life anxiety symptoms are common and associated with increased impairment, distress, health care utilization, and risk for developing psychiatric disorders.⁹ Anxiety disorders are independently associated with several physical conditions which in turn, are associated with poor quality of life and disability,¹⁰ and in comparison with depression-only elderly patients, the co-occurrence of depression and generalized anxiety disorder represents more severe and chronic psychopathology.¹¹

Our goal was to examine the characteristics of older patients identified by their primary care physicians as anxious, with a focus on the ethnicity of the patient. Although depression and anxiety often co-exist,¹¹ we are looking for an independent effect for anxiety even after adjusting for depression symptoms and other patient variables thought to be associated with identification. Our conceptual model incorporated severity of anxiety symptoms, functional status, medical conditions, and physician’s ratings how well they knew the patient as possible influential covariates in the relationship between patient ethnicity and identification of anxiety by primary care physicians. We acknowledge that race and ethnic origin are crude markers of complex social and behavioral patterns and that designations of ethnicity imply a homogeneity of groups, which is an oversimplification. Ethnicity refers to a common heritage shared by a particular group¹² and consistent with the National Institutes of Health and current research, we use the terms black to include individuals of African, African American, and African Caribbean descent and white to include individuals of European descent. We hypothesized that black patients might have a different experience regarding anxiety identification and use of psychotropic medications compared to white patients.

METHODS

The Spectrum Survey

The Spectrum Survey was an observational study among older primary care patients which included both primary care physician and patient assessments. The study has been described in detail elsewhere.¹³ In summary, primary care practices recruited from the community provided the venue for sampling older patients. Participants who agreed to be part of the study were scheduled for an in-home interview which consisted of a 90 minute survey questionnaire. In-home interviews were obtained for 357 people, but 2 persons broke off the interview before it was completed, leaving a sample of 355 persons. The study protocols were approved by the Institutional Review Board of the University of Pennsylvania School of Medicine.

Physician Assessment of Anxiety

Physicians were asked to provide their assessment of patient’s anxiety at the index visit. Physicians were asked to rate the patient’s level of anxiety 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 anxiety. Physicians were asked “How well do you know this patient?” and were asked to choose among the following response categories “very well, somewhat, or not at all.” Physicians rated the patient’s focus on medically unexplained somatic complaints on a 5-point scale: “none, a little, somewhat, significant, or a great deal.”

Patient Assessment

We used standard questions to obtain information from the respondents on age, gender, marital status, self-reported ethnicity, and education. 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. Patients were asked “During the past 6 months, how many visits did you make to the primary care or family doctors, internists, surgeons or other medical specialists? This question refers only to office visits or clinic visits.” Questions from the 36-item Short-Form Health Survey (SF-36) were used to assess functional status.¹⁴ Baseline medical comorbidity was measured by summing the lifetime presence of chronic diseases or conditions. The Centers for Epidemiologic Studies Depression (CES-D) scale was developed by the Center for Epidemiologic Studies at the National Institute of Mental Health for use in studies of depression in community samples.¹⁵ The Beck Anxiety Inventory (BAI) is a 21-item self-report instrument developed in order to measure the severity of anxiety symptoms and has been shown to be an appropriate instrument for measuring symptoms of anxiety in the elderly.¹⁶ Total scores range from 0 to 63. Beck et al. suggest that scores of 15 or greater represent moderate to severe levels of anxiety. The Beck Hopelessness Scale (BHS) consists of 20 statements, rated by the respondent as true or false, and measures negative expectations about the future, a sense of giving up, and future anticipation or plans.¹⁷ Participants were asked to gather the bottles of their prescription medications before the interview. The interviewer wrote down the medications and dosages directly from the bottles.

Analytic Strategy

Data analysis was performed using SPSS version 12 and SAS version 9. We compared characteristics of participants who self-identified as black with those of participants who self-identified themselves as white, using χ² or two-tailed t-tests as appropriate for categorical or continuous data. Multivariate models with random effects for practice were used to account for clustering due to practice. Our primary dependent variable was identification of the patient as anxious by the primary care physician. We also examined identification in relation to whether or not the participants scored 15 or higher on the BAI.¹⁶ Our study sample included 355 participants who had completed a baseline in-home interview. Twenty-five participants were excluded because of incomplete primary care physician assessment or self-identified ethnicity, leaving a sample size of 330 for this analysis. We employed separate multivariate logistic regression models to assess the relationship of ethnicity to physician identification of anxiety. These multivariate models were adjusted for sociodemographic factors (age, gender, level of educational attainment, and martial status), functional status, physical health, and adjusted for both depression and anxiety symptoms. A subsequent model included terms for familiarity of the practice, as represented by the number of visits made to that practice in the past 6 months before the interview, and the physician’s ratings of how well they knew the patient. We also tabulated the baseline psychotropic medication use among patients identified as anxious with the purpose of comparing differences in medication usage by patient ethnicity. The odds ratio was our measure of association.

RESULTS

Study Sample

The mean age ± standard deviation of our study sample was 75.3 ± 5.9 years with a range of 65 to 92 years. Of the participants, 248 (75.2%) were women. The self-identified ethnic groups consisted of 217 (65.8%) white and 113 (34.2%) black patients. Proportionately fewer black patients were identified by their primary care physicians as anxious (p<.001).

Physician Identification of Anxiety

Black patients who were identified as anxious, when compared to white patients identified as anxious, were more likely to report worse social functioning and bodily pain (Table 1). Although anxious white patients reported a greater number of visits to their physician’s office in the past 6 months, physicians indicated that they knew anxious black patients as well as anxious white patients. Anxious black patients had more depression symptoms, poorer emotional functioning, more bodily pain, and focused more on medically unexplained somatic complaints than black patients who were not identified as anxious. Anxious white patients had more depression, anxiety, and hopelessness symptoms, poorer physical and emotional functioning, and focused more on medically unexplained somatic complaints than white patients who were not identified as anxious. Black patients who were not identified as anxious had poorer physical functioning than white patients who were not identified as anxious.

Table 1.

Comparison of Patients Identified by Primary Care Physicians as “Not at All” Anxious With Patients Identified as Anxious.

Variable	“Not at All” Anxious		Anxious
	Black Patients (n=60)	White Patients (n=60)	Black Patients (n=53)	White Patients (n=157)
Demographics
Age, y	74.6 (5.6)	76.2 (6.2)	75.5 (5.3)	75.2 (6.1)
Women, No. (%)	46 (76.7%)†ψ	34 (56.7%)†§	49 (92.5%)‡ψ	119 (75.8%)‡§
Education less than high school, No. (%)	40 (66.7%)†	12 (20.0%)†§	27 (50.9%)‡	53 (33.8%)‡§
Married, No. (%)	15 (25.0%)†	33 (55.0%)†	13 (24.5%)‡	69 (43.9%)‡
Psychological
CES-D Score (0–60)	11.9 (9.5)ψ	10.6 (9.6)§	19.2 (12.5)ψ	15.6 (11.2)§
BAI Score (0–63)	7.8 (7.8)	6.6 (5.7)§	10.4 (8.3)	9.6 (8.8)§
BHS Score (0–20)	3.9 (3.3)	3.9 (3.0)§	5.2 (4.6)	5.4 (4.2)§
Physical Health
Physical function score (0–100)	55.6 (31.4)†	67.9 (26.9)†§	51.4 (29.6)	58.6 (28.0)§
Role physical score (0–100)	36.7 (40.3)†	51.5 (39.5)†	38.2 (42.6)	45.9 (38.3)
Role emotional score (0–100)	81.1 (35.5)ψ	88.9 (29.2)§	62.3 (43.9)ψ	71.7 (39.8)§
Social function score (0–100)	72.5 (27.6)	75.4 (25.8)	62.7 (32.8)‡	72.2 (25.7)‡
Bodily pain score (0–100)	57.2 (23.1)ψ	56.1 (21.9)	44.4 (29.4)ψ‡	54.6 (25.0)‡
General health perception score (0–100)	53.3 (19.9)	54.6 (17.4)	47.3 (20.4)	50.0 (20.8)
No. of medical conditions (0–27)	7.3 (3.0)	6.5 (2.8)	7.9 (3.7)	7.3 (3.8)
No. of visits to office within 6 mo.	3.0 (1.7)	3.7 (2.7)	2.9 (1.9)‡	4.5 (4.7)‡
Patient’s focus on medically unexplained somatic complaints: “a little, somewhat, significant, a great deal,” No. (%)	25 (41.7%)ψ	21 (35.0%)§	44 (83.0%)ψ	123 (79.4%)§
Physician ratings at index visit, No. (%)
Knows the patient “very well”	40 (66.7%)	44 (73.3%)	40 (75.5%)	126 (80.3%)

Note: Data were gathered from the Spectrum Survey, 2001–2003. p-Values given for comparison of groups with chi-square, t-test, or ANOVA, as appropriate. Percentages are column percentages. Ranges are shown in parentheses.

Abbreviations: BAI = Beck Anxiety Inventory; CES-D = Center for Epidemiological Studies Depression Scale; BHS = Beck Hopelessness Scale

^†p<.05 for comparison of patients identified as not anxious.

^‡p<.05 for comparison of patients identified as anxious.

^ψp<.05 for comparison of black patients who were identified as not anxious with black patients who were identified as anxious.

^§for comparison of white patients who were identified as not anxious with white patients who were identified as anxious.

Multivariate Models of Identification

The association of patient ethnicity and primary care physician identification of anxiety was evaluated by using multiple logistic regression models with random effects for practice. In the unadjusted model, black patients were about a third as likely to have been identified as anxious than white patients (Odds Ratio (OR) = 0.34; 95% Confidence Interval (CI) [0.18, 0.64]). Additional multivariate analyses were performed that adjusted for sociodemographics, functional status, psychological status, medical conditions, number of office visits in the past 6 months, and the physician’s rating of how well they knew the patient in a step-wise fashion. In both models (Model 1 which adjusted for sociodemographics, functional status, psychological status, medical conditions, number of office visits in the past 6 months; and Model 2 which adjusted for number of office visits in the past 6 months and the physician’s ratings of how well they knew the patient, in addition to all the terms included in Model 1), the association between patient ethnicity and identification of anxiety remained unchanged; (OR = 0.28; 95% CI [0.14, 0.58]) and (OR = 0.30; 95% CI [0.15, 0.61]), respectively. Among persons who scored above the threshold on the BAI, black patients were also about a third as likely to have been identified as anxious than white patients (unadjusted OR = 0.32; 95% CI [0.09, 1.00]).

Psychotropic Medication Use

Baseline psychotropic medication use among patients identified as anxious was examined. Anxious black patients compared to anxious white patients were less likely to be taking a selective serotonin reuptake inhibitor (SSRI) (20.8% versus 35.7%; p=0.047; OR = 0.47; 95% CI [0.23, 0.99]) or a benzodiazepine (3.8% versus 14.0%; p=0.060; OR = 0.24; 95% CI [0.06, 1.06]). The results for benzodiazepines approached but did not reach statistical significance. Only a small number of patients were taking a serotonin norepinephrine reuptake inhibitor (SNRI), a tricyclic antidepressant (TCA), or buspirone. In all, 2 anxious white patients (1.3%) while no anxious black patients were taking a serotonin norepinephrine reuptake inhibitor (SNRI), 8 anxious white patients (5.1%) compared to 2 anxious black patients (3.8%) were taking a tricyclic antidepressant (TCA), and 2 anxious white patients (1.3%) compared to 1 anxious black patient (1.9%) were taking buspirone. Anxious black patients compared to anxious white patients were significantly less likely to be taking any psychotropic medication (24.5% versus 44.6%; p<0.011; OR = 0.40; 95% CI [0.20, 0.81]).

DISCUSSION

In this community-based primary care sample, older black patients were less likely than older white patients to be identified as anxious and were less likely to be taking psychotropic medications, selective serotonin reuptake inhibitors and benzodiazepines. The association between patient ethnicity and physician identification of anxiety persisted even upon controlling for potentially influential variables, including severity of anxiety symptoms, functional status, medical conditions, and physician’s ratings how well they knew the patient. Our study contributes to previous work examining the role of ethnicity in the identification and management of mental illness,^{1, 3, 4, 18–21} and suggests that patient ethnicity may play a role in the identification and management of anxiety among older adults who present to their primary care physicians.

Before discussing our findings, the results must first be considered in the context of some potential study limitations. First, we obtained our results only from primary care sites in Maryland whose patients may not be representative of most primary care practices. However, these practices were not academically affiliated and are probably similar to other practices in the country. Second, there is the potential for the sources of error associated with retrospective interview data including imperfect recall and response bias (i.e. socially desirable responding). Third, the survey instruments used in the study do not necessarily reflect the actual interaction the patient had with the physician during when the patient was assessed by the physician. We do not know if there are ethnic differences in symptom expression during the doctor-patient encounter and we do not have supplemental ratings of patient behavior or symptoms from other healthcare workers.

Nevertheless, despite limitations, our study warrants attention because we attempted to address the relationship between patient ethnicity and identification of anxiety while adjusting our estimates of association for demographic factors, functional impairment, psychopathology, medical conditions, and the physicians’ ratings of how well they knew the patient. In addition, because we examined a community-based primary care sample, the results may be generalizable to older adults in primary care settings, which may help facilitate identification and treatment of anxiety in the future. Furthermore, only a few studies have focused on older adults to assess the relationship between physician ratings, patient characteristics, and identification of mental illness, with most of these studies being limited to depression. For example, Gallo et al found that black patients 65 years and older were less likely to be identified as depressed and their depression was less likely to be actively managed than older white patients.³

The differences in psychotropic medication use according to ethnicity are disconcerting, as there were no significant differences in depression symptoms and also in the distribution of severity of anxiety between anxious black and anxious white patients. Selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs), in conjunction with cognitive-behavioral therapy, are considered first-line interventions for anxiety, due to their efficacy and relative tolerability in elderly patients.²² The greater proportion of anxious white patients on SSRIs is consistent with past research documenting that the increase in antidepressant use among older patients has occurred mostly in white patients;⁴ that black patients are less compliant with depression treatment regimens, even though no ethnicity-related treatment response difference has been identified;⁵ and that black patients are less likely than white patients to find antidepressant medication acceptable.²⁰ The high proportion of elderly patients (especially white patients) on benzodiazepines is worrisome, as the negative side effect profile and toxicity of benzodiazepines in older adults has been well established. Problems associated with benzodiazepine use by the elderly include sleep disturbances, cognitive difficulties, impairment in activities of daily living, motor vehicle crashes, and accidental falls and fall-related fractures.^23–28 Despite guidelines defining the appropriate use of benzodiazepines in the elderly (for prescriptions to be brief, intermittent, and for purposes of acute symptom relief), benzodiazepine use in older adults remains high with a mean current prevalence of 12.3% in community-dwelling populations.^{29, 30}

Our study emphasizes that physicians caring for black patients should be cognizant of the tendency for anxiety symptoms to be judged not as severe. In order to facilitate the doctor-patient interaction, physicians might introduce this fact into their interviews when appropriate and determine whether this leads to a different conversation with their black patient. Acknowledging that possibility of a discrepancy occurs may be the first step toward improving the doctor-patient communication and ultimately, the identification and management of anxiety. We acknowledge that the recognition and management of anxiety in primary care settings is a complex process between the patient and the primary care physician that is influenced by multiple provider, patient, and practice-setting attributes. Further research is essential to clarify how the characteristics of the physician, patient, and health care systems influence the doctor-patient interaction.