Race, Stigma, and Mental Health Referrals Among Clients of Aging Services Who Screened Positive for Depression

The U.S. Department of Health and Human Services’ Administration on Aging (one of three agencies that make up the Administration for Community Living) authorizes meal provision to individuals who are age 60 or over and homebound, providing home delivery of 149 million meals to more than 880,000 individuals in 2009. Compared with the overall U.S. population aged 60 and older, these meal recipients were more likely to be older, poor, and black; to live alone; to be in poor health; to have greater difficulty performing everyday tasks; and to be at high nutritional risk (11). To integrate mental health assessment into aging services, the Westchester County Department of Senior Programs and Services entered into a community-academic partnership with the Weill Cornell Institute of Geriatric Psychiatry, part of the Department of Psychiatry at Weill Cornell Medical College, and depression screening was added to the routine assessment of all applicants for home meal service.

To integrate screening for mental health treatment into routine assessments, the community-academic partnership chose measures through a collaborative and iterative process. All consecutive applicants for home meal service undergo a routine assessment of functioning, nutritional risk, and social support that culminates in recommendations for nutritional support and other services. Supplemental measures that have low client burden, are easy to administer by staff without a mental health background, and can produce service recommendations were selected.

The Patient Health Questionnaire (PHQ-9), a widely used depression screen, was added to detect clinically significant depressive symptoms (score ≥10) that would warrant a mental health referral (12). The Blessed Memory Orientation and Concentration (BMOC) test was used to identify cognitive impairment (score ≥16) (13). Finally, the anticipated cost of stigma (ACS) scale was administered to capture concerns that disclosure of depression or mental health treatment would result in negative treatment by people in one’s social network. The ACS scale contains seven items describing negative reactions that a respondent might expect from friends and family if he or she was depressed, for example, “exclude me.” Responses to each item range from 1, strongly disagree, to 4, strongly agree. [The ACS scale is available online as a data supplement to this report.]

A median split of scores was used to identify individuals who were most concerned about anticipated stigma. Scores above and below the median were considered indicators of high and low anticipated stigma, respectively. In previous analyses with diverse, English-speaking samples, the scale was found to have good internal consistency (Cronbach’s α=.89) (14). All of the selected measures have been used with older adults in previous research studies.

To improve the detection of depression and referral for treatment, the academic partner provided staff training on the administration and scoring of the supplemental measures. Staff were trained to make a referral for community mental health resources for clients with scores ≥10 on the PHQ-9 (the standard cutoff score). Referral guidelines, when combined with screening, increase the likelihood of further assessment in other care settings. The routine assessment, administration of supplemental measures, and provision of the mental health referral were completed in one visit. A written referral form to indicate the services recommended was completed. This written referral was defined as a successful referral. This secondary data analysis project was approved by the Weill Cornell Internal Review Board. Data were collected consecutively on new meal program enrollees from December 2004 until June 2006.

Data analyses were conducted on the deidentified data for the total sample screened and the subgroup of individuals who had clinically significant depression. Bivariate comparisons (t and chi square tests) were conducted to examine racial differences in anticipated stigma between African-American and Caucasian older adults in the total sample. In the subgroup of depressed clients, racial differences in anticipated stigma were reexamined. To examine the relationship between high anticipated stigma and a referral for mental health treatment among depressed clients, we developed a parsimonious linear model by using a logistic regression analysis that included the presence or absence of a successful mental health referral as the outcome variable. The final model was subjected to the Hosmer-Lemeshow goodness-of-fit test. All data analyses were done by using SPSS, version 19.0.

The total sample comprised 732 adults, including 540 (74%) women; 190 (26%) self-identified as African American and 542 (74%) self-identified as Caucasian, with 16 Caucasians indicating that they were of Hispanic origin. The clients’ mean±SD age was 82±8.8 years, and they had 12.5±3.1 years of education. Only education level differed by race, with Caucasians having more years of education than African Americans (12.7±3.0 versus 11.7±3.4). Clients with missing data constituted less than 1% of the sample and did not differ from the larger sample.

Case management (additional follow-up visits) was recommended to 89 clients (12%), and 46 (6%) clients were referred for additional nutritional counseling. Only a fraction of the sample (N=20, 3%) received a referral for both nutritional counseling and case management. Compared with Caucasians, African Americans were referred significantly more often for nutritional counseling (N=24 of 190, 13%, versus N=22 of 542, 4%; χ²=17.61, df=1, p<.001).

A total of 81 (11%) clients endorsed significant depressive symptoms, with similar rates among African Americans (N=22, 12%) and Caucasians (N=59, 11%). A subgroup of clients (N=70, 10%) had cognitive impairment on the BMOC test, with African Americans experiencing twice the rate of cognitive impairment as Caucasians (N=31, 16%, versus N=39, 7%; χ²=14.3, df=1, p<.001).

A successful mental health referral was documented for 43 (6%) clients, with African Americans (N=11, 6%) and Caucasians (N=32, 6%) receiving referrals at the same rate. Most (N=31, 72%) referrals for mental health treatment were for depression. Cognitive impairment, age, and years of education were unrelated to being offered a mental health referral. Clients who were referred for mental health treatment were more likely than those who were not referred for mental health treatment to receive a case management referral (N=13 of 43, 30%, versus N=76 of 689, 11%; χ²=5.11, df=2, p<.05).

The mean score on the ACS scale was 2.09±.51. African Americans endorsed high stigma more frequently than Caucasians (N=130, 68%, versus N=288, 53%; χ²=13.66, df=2, p<.001). Rates of high anticipated stigma varied significantly by race and gender group (χ²=14.42, df=3, p<.01), with African-American men most likely to endorse high anticipated stigma (N=46 of 63, 73%).]

The subgroup with depression was significantly more likely than the group without depression to endorse high anticipated stigma (N=57, 70%, versus N=360, 55%; χ²=7.32, df=1, p<.01). Of the 81 clients with depression, 31 (38%) successfully received a mental health referral. A logistic regression was conducted to examine the relationship of race, gender, age, depression severity (PHQ-9 score), and anticipated stigma and a successful mental health referral among clients with depression. To capture potential gender × race differences, an interaction term was created. In the final model, low anticipated stigma, younger age, and race × gender interaction significantly (p<.05) predicted successful receipt of a mental health referral among clients with depression. In this subsample, there were no significant associations between receiving a mental health referral and cognitive functioning or depression severity. The relationship between receipt of a mental health referral and a recommendation for case management at the bivariate level was not sustained in the final model. The maximum likelihood estimates of the main and interaction effects are presented in Table 1.

Age was associated with receiving a mental health referral, such that for each year of age the odds of not receiving a referral (versus receiving a referral) increased by a factor of 1.067 (=1/exp[–.065]). The odds of getting a mental health referral were 70% lower among adults with high versus low anticipated stigma. The interpretation of the significant interaction term between gender and race is not straightforward and can only be described by taking into account the nonsignificant main effects for these variables. Compared with odds for an African-American woman, the odds that an African-American man would get a referral were .26 (=exp[−3.96*.067]). Keeping all other factors constant, a man of African descent had the least possible chance of getting a mental health referral, whereas a Caucasian woman had the highest chance of a successful referral.

The primary findings of this study were that anticipated stigma varied by race and gender and that adults with depression who reported high anticipated stigma were less likely than those who reported low anticipated stigma to receive a referral. Among clients who were not depressed, African-American men and women had the highest scores for anticipated stigma. They were more likely to express concern that being depressed or seeking mental health treatment would result in being negatively treated by their social network. Alarmingly, among depressed adults the likelihood of being successfully referred for mental health treatment was associated with the level of anticipated stigma. Specifically, clients who were younger and had lower anticipated stigma were more likely to be successfully offered a referral after detection of depression. African-American men were the least likely to be referred for mental health care.

The finding of the interrelation of gender, race, anticipated stigma, and a successful mental health referral illustrates the compounded barriers to mental health care for older adults from racial-ethnic minority groups. Our findings add to research documenting individual-level barriers reported by older members of minority groups in other non–mental-health settings. African Americans in a primary care setting were concerned about the public stigma from family, friends, and their community and they viewed seeking mental health services as a form of weakness. Asking about anticipated stigma also highlighted the role of social networks in obtaining mental health treatment. As increasing attention is paid to barriers, awareness of the role of a client’s social network may improve the likelihood of a successful referral for mental health treatment.

A limitation of this study was that it captured a single moment in a complex process of being referred for mental health treatment. The data did not capture the subtle interactions between clients and staff that could elucidate how anticipated stigma, gender, race, and age affect the mental health referral process. Staff may offer mental health referrals later or only when they believe that a referral is likely to be accepted, even when referral guidelines are clear. Because this work was the product of a community partnership, the factors available to predict the referral process were limited. Other client characteristics, such as income, functional status, or medical burden, may have affected who received successful referrals. Likewise, the characteristics of the staff and their beliefs were not assessed and could have affected the client outcomes. In addition, our measure of anticipated stigma has been used in research with older adults in other settings, such as primary care and home care, but has not been validated cross-culturally. Finally, this sample of older adults did not represent the heterogeneity of the African-American or Caucasian older adult population, but it was consistent with the national profile of participants in the homebound Nutrition Services Program (www.hhs.gov/asl/testify/2011/06/t20110621a.html). Future work should include other racial and ethnic groups served by the ASN, including a larger sample of Latino elders.

Aging services can provide a unique opportunity to detect depression, but screening cannot be implemented effectively without addressing the challenges of referring clients with depression for mental health treatment and keeping them engaged (15). The challenges exist at both the individual and systemic levels. Anticipated stigma may affect both the older person’s willingness to consider a mental health referral and the staff’s ability to offer one, even when training for screening and referral resources are adequate. In the move to the delivery of mental health services to older adults, we need to consider the complexity of detection and referral in non–mental-health community settings. As mental health care for older adults increasingly falls “into the hands” of the aging services, interventions that address barriers may facilitate successful referrals.

Support for this research was provided by grants R01 MH 079265 and P30 MH68638 from the National Institute of Mental Health. The authors appreciate the support of the Westchester County Department of Senior Programs and Services, Commissioner Mae Carpenter, M.S., Deputy Commissioner Diane Aparisio, M.S.W., and all the seniors who participate in the home meal programs.

The authors report no competing interests.