National studies have found that persons from minority racial-ethnic groups are less likely than non-Hispanic whites to receive services for mental disorders (1). A recent analysis of national trends showed little abatement in these disparities (2). A primary research-to-practice goal of disparity research is to target remediation strategies. Although national data are informative, the use of sampling strategies to achieve broad coverage and survey requirements for broad categories of data collection prevent detection of local patterns of racial-ethnic differences, limiting the usefulness of national data sets for formulating disparity remediation strategies.

Heterogeneity in local patterns of service use by racial-ethnic groups is to be expected because of variation across the country in the demographic composition of racial-ethnic groups and in the organization of local mental health care systems. For example, Hispanic communities in the Southwest contain greater proportions of Mexican Americans than do communities in the Northeast, where the proportion of Puerto Rican Americans is higher (3). Because Mexican Americans have markedly lower rates of use of specialty mental health services than Puerto Rican Americans (4), priorities for disparity reduction in Southwestern communities may differ considerably from those in the Northeast. Local variation in service system configurations, including the geographical distribution of providers across states, urban versus rural location, regional variation in provider practices (5), and the cultural competency of available services, may also affect racial-ethnic patterns of use. Far from being of local interest only, studies in defined contextual settings permit a more critical appraisal and understanding of the reasons for disparities. As more results of such intensive studies become available, the greater will be our comprehension of the national scene and the viability of various approaches to reduce disparities.

We report on a study of differences in the receipt of mental health services among blacks, Hispanics, Asians, and whites in New York, a large culturally diverse state. We used comprehensive state data for all persons, except inpatients, receiving state-regulated services, regardless of payer. We examined treatment in programs ranging from acute clinical care to tertiary, recovery-oriented services. The service users were clinically heterogeneous and included persons with psychoses and other chronic illnesses, a group we note is not well represented in national epidemiological surveys. We conducted detailed analyses by service type, diagnosis, and population density to allow specific recommendations to be made for disparity reduction.

The null hypotheses tested were that annual rates of treated prevalence and treatment intensity rates would not differ between whites and members of each group. Further, we investigated whether area demography affects these rates.

Data

Data were drawn from the Patient Characteristics Survey (PCS) (6) conducted by the New York State Office of Mental Health (NYS OMH), a biannual survey on mental health care service utilization in all state-regulated programs, defined as those operated, funded, or licensed by NYS OMH. Persons are eligible to use these services independent of income or receipt of public entitlements. The survey excludes private providers who deliver services outside of these programs; mental health visits to general hospital emergency departments, because they are not regulated by OMH; and services delivered in environments that are exclusively federally funded. The survey is conducted in a “typical” week in the year, meaning that no unusual service scheduling problems (such as holidays) are expected. For each person attending a program in that week, information is collected on the types of services received in the week and on basic demographic and clinical characteristics. Data are obtained from program records and hence reflect the approaches used in these programs to collect data. Race and ethnicity are client self-assessments, and diagnoses are assessed by program staff. We used 2009 PCS data for non-inpatient users (excluding those in residential treatment). Prevalence denominators were 2009 updated census estimates of NYS resident population (7). The survey data are deidentified, and because the study was neither clinical research nor a clinical trial, it received exempt status from the institutional review board.

Access measures, likelihood of service use, and treatment intensity

The time since each client’s receipt of last service in the program was used to estimate for each racial-ethnic group the unduplicated number served in the year (8) and the average annual number of weeks in which at least one service visit was made per person served (9). The latter estimate was considered a measure of treatment intensity. The proportion of clients using a service type was used as a measure of the likelihood of service use. Access was measured by the annual treated prevalence, obtained by dividing the estimate of the number of unique individuals who received any mental health services in the year by the number in the group population. The denominator was a count of all persons in the population because services were available to all. The estimators were asymptotically unbiased, which follows from the large sample size and the assumption that the survey was conducted in a typical week.

Categories

Race-ethnicity was grouped into white non-Hispanics, black non-Hispanics, Hispanics, and Asians. Pacific Islanders were few and were included only in prevalence rate analyses because census data used in prevalence denominators aggregated Asians and Pacific Islanders. There were too few Native Americans for analyses. Persons who self-reported multiple racial categories were not included because biracial categories were numerous, each containing too few persons for analyses. Ages were grouped for youths three to 17, adults 18 to 65, and adults over 65. Primary DSM-IV diagnoses were grouped for youths into attention-deficit hyperactivity disorder (ADHD) and mood, disruptive behavior, adjustment, and anxiety disorders; for adult age groups, diagnostic groups included psychoses, depression, bipolar disorders, and anxiety disorders. No groupings for comorbidities were used.

Service type groupings were emergency services, comprising mainly OMH-sponsored psychiatric emergency rooms and brief-stay crisis intervention services; clinic services, comprising behavioral and group therapy visits as well as visits for only medication treatment; other outpatient services, including day treatment, assertive community treatment, continuing day treatment, partial hospitalization, rehabilitation, and recovery-oriented services; and community support, comprising services from psychosocial clubs (settings that provide supports and opportunities to learn life skills and vocational competencies to be self-sufficient in society), case management, assisted competitive and supported employment, advocacy and support services, drop-in centers, and care coordination. Days in community residences were not counted. Regions were grouped into counties by geographical location and population density into the New York City (NYC) metropolitan area, covering the NYC counties as well as large neighboring suburban counties of Westchester, Rockland, Nassau, and Suffolk; upstate metropolitan areas, covering upstate cities with populations >50,000; and upstate nonmetropolitan areas, covering smaller towns and rural areas. In analyses, persons were assigned to their county of residence.

Statistical analysis

The null hypotheses of no pairwise differences in treated prevalence rates and in treatment intensity rates between whites and another racial-ethnic group were tested with methods developed by Laska and colleagues (8,9). Tables note all significant differences in prevalence and significant differences of two or more weeks of service use. Chi square tests were performed to test equality of the distributions of client characteristics and the likelihood of use of a service type among racial-ethnic age groups. Because of the large sample sizes, all chi square tests were significant at the .05 level. We comment only on differences in percentages of 5% or more. No adjustments were made for multiple comparisons.

Study population characteristics

Table 1 presents demographic, diagnostic, and insurance information on the racial-ethnic population studied. A total of 142,959 NYS residents were served in the week in 2,500 non-inpatient programs, resulting in an estimate of 578,496 persons served in the year. Of the annualized population, 49% were white, 24% black, 25% Hispanic, and 2% Asian. In 2010, the corresponding population figures for the racial-ethnic groups were, respectively, 54%, 18%, 20%, and 8%. In the annualized population, 21% were youths, 73% adults through age 65, and 6% adults over 65. Blacks and Hispanics had slightly higher proportions of youths and lower proportions of older adults for the studied services.

ADHD was among the two most prevalent diagnoses for all youth groups. In addition, among youths a prevalent diagnosis for whites was mood disorders; for blacks, disruptive behaviors; and for Hispanics and Asians, adjustment disorders. Among the adult age groups, whites and Hispanics were most likely to have a depressive disorder diagnosis, whereas blacks were considerably more likely than whites and Hispanics to be diagnosed as having a psychotic disorder. Asian adults ≤65 were most likely to have diagnoses of psychotic disorders, with depressive disorder most prevalent among the older Asians. White youths were more likely than other groups to be covered by private insurance, whereas Hispanic, black, and Asian youths were more likely to have public insurance. Asian and white adults ≤65 had similar insurance patterns, as did Hispanics and blacks in the same age group. White adults ≤65 were more likely to have private insurance than blacks and Hispanics, but at half the percentage of white youths (19% versus 38%). Substantial percentages of all racial-ethnic adult groups ≤65 reported having no insurance (14%–18%), with the highest rate reported for blacks. For all older adults, 84% or more had public insurance.

Statewide and regional prevalence and treatment intensity

Table 2 presents racial-ethnic differences in treated prevalence and treatment intensity by persons’ age and region. Statewide for all age groups, blacks and Hispanics had treated prevalence rates 1.5 to 3.0 times higher than whites, whereas rates among Asians were 1.5 to 3.0 times lower. Treated prevalence for both young and adult whites ≤65 increased as areas become more rural. Rates among black and Hispanic youths and adults ≤65 were highest in upstate metropolitan areas. In upstate nonmetropolitan areas, black and Hispanic youth rates did not statistically differ from those for whites, whereas rates for black and Hispanic adults ≤65 were higher than those for whites.

Across all age groups statewide, treatment intensity rates among black, Hispanic, and white groups ranged from 12 to 14 weeks. Statewide, Asian youths and adults ≤65 had rates comparable with those of whites, but Asians >65 had substantially lower rates (eight weeks versus 13 weeks). Compared with their white counterparts, black youths in upstate metropolitan areas received services in approximately two fewer weeks, and blacks >65 in the NYC metropolitan area received services in three fewer weeks. Hispanic adults ≤65 in upstate metropolitan areas received services in two fewer weeks than whites.

Statewide likelihood of service use and treatment intensity

Tables 3 and 4 display statewide likelihoods of service type use and treatment intensity rates for specific service types among youths (Table 3) and adults ≤65 and >65 (Table 4), stratified by diagnosis. Black and white youths had similar patterns of use for specific service types. Black adults ≤65 and >65 were less likely than white counterparts to have received a clinic service (47% versus 58% and 50% versus 63%, respectively) and more likely to have received community supports (34% versus 28% and 39% versus 25%, respectively). Black adults ≤65 with psychoses were more likely than whites to have received an emergency service (29% versus 17%).

Hispanic youths for all diagnoses in comparison with whites were more likely to have received a clinic visit (74% versus 63%) and less likely to have received community supports (12% versus 22%). The pattern was the same for Hispanic adults. Compared with whites, younger and older Hispanic adults were more likely to receive a clinic visit (65%–71% versus 58%–63%, respectively) but less likely to receive community supports (20%–23% versus 25%–28%, respectively). Younger and older Hispanic adults with psychoses were more likely than whites to have received an emergency service (ages ≤65, 25% versus 17% and ages >65, 47% versus 18%, respectively). Asian youths were more likely than whites to use clinic services (81% versus 63%) and less likely to use community supports (5% versus 22%). Asian youths with ADHD were more likely than whites to receive an emergency service (33% versus 5%). Compared with whites, Asian adults ≤65 were more likely to have received other outpatient services (13% versus 8%) and the older Asian adults were more likely to have received emergency services (34% versus 14%).

Black youths received community supports in five fewer weeks than whites, and black youths with disruptive behaviors received community supports in approximately half the number of weeks as whites (11 versus 21 weeks). Black adults ≤65 with bipolar disorders in contrast to white counterparts received community support services in four fewer weeks. Hispanic youths with anxiety disorders received clinic services in six more weeks compared with whites, whereas Hispanic adults ≤65 and >65 with anxiety disorders received, respectively, three and seven fewer weeks of clinic services than white counterparts. Hispanic adults ≤65 with depressive disorders did not differ from whites in the number of weeks of clinic services, but the older Hispanic adults had clinic services in two fewer weeks than their white counterparts. Across all diagnoses, Asian youths and adults ≤65 using treatment did so at rates similar to or higher than those of whites: Asian youths received community supports in more weeks (41 versus 17 weeks), specifically youths with mood disorders (38 versus 18 weeks). Asian adults ≤65 with bipolar disorder received higher rates of community supports (30 versus 18 weeks) and emergency services (four weeks versus two weeks).

Access

In contrast to national findings, treated prevalence was greater for Hispanics and blacks than for whites. This may be explained by the fact that the NYS PCS survey does not cover private-sector services delivered by solo private practitioners, whereas national findings are based on community surveys in which persons who screen positive for mental disorders report on their access to behavioral health care services in both the public and private sectors (10). The lower treated prevalence rates of whites compared with Hispanics and blacks in the state-regulated mental health system are likely offset by whites’ greater use of private providers, because, on average, whites have higher incomes (11) and are more likely to have private insurance (12). This conjecture is supported by the finding that treated prevalence for white adults and youths was higher in rural than urban areas, which may be due to a smaller supply of private providers (13) and to lower average income levels among whites in rural areas. The lower treated prevalence rate for Asians is directionally in agreement with national findings. Similar to whites, Asians with higher incomes may seek care from private mental health providers. But many Asians may receive care from primary care physicians because of a cultural preference or the likelihood of somatic presentation of symptoms in accordance with the culture’s holistic view of mind and body (14). Stigma and family shame experienced by Asians may also play a particularly large role, lessening the likelihood that they will seek services (15). The elderly and less acculturated Asians with limited English proficiency may have little contact with community agencies that might refer them to needed services.

Treatment intensity

For persons who access treatment, both national and local study findings on racial-ethnic differences vary, from no evidence of disparities (16) to significant disparities (17). In New York, statewide aggregated treatment intensity rates did not substantially differ among racial-ethnic groups and age groups—with the exception of Asian elderly adults, who had substantially lower treatment intensity rates. More detailed analyses, however, revealed noteworthy differences.

Blacks.

Although Hispanic and Asian youths across diagnostic groups did not differ from or exceed whites in treatment intensity, black youths received community support services in fewer weeks compared with white youths, particularly among youths with disruptive behaviors. Poverty-related community stressors leading to drug and alcohol use and involvement with the juvenile justice system may impede their use of services. Differences were noted in blacks’ rates between upstate and downstate metropolitan areas, which may reflect differential accessibility between areas to culturally competent providers whose importance to black youths has been noted (18,19). These youths may be deterred from remaining in services because they may view providers as lacking an understanding of their cultural context and being nonwelcoming. Compared with whites, black adults ≤65 with psychoses used emergency services more and used clinic services less. Increasing the cultural competency of clinic services for this group could work to circumvent their greater use of emergency services, but greater insight is still required as to how this might be accomplished. Black older adults in the NYC metropolitan area who received services in fewer weeks compared with white counterparts may have had limited access to caregivers who could facilitate their transportation to keep appointments. For older black adults, caregivers are often younger female family members with extensive child care obligations and daytime work schedules (20,21).

Hispanics.

For Hispanics, statewide treatment intensity rates for youths did not differ from whites. Schools and other child-related systems in NYS may be successfully working to engage Hispanic youths. But adults with anxiety disorders and older adults with either a depressive disorder or an anxiety disorder had differential rates of clinic care. Diagnostic issues (discussed below) may play a role. All Hispanic adults with psychoses were more likely than white counterparts to use emergency services and less likely to use clinic care. Stigma might be a barrier to receiving ongoing clinic care (22). Culturally competent engagement and retention strategies for Hispanic adults may be required to increase their use of clinic services.

Asians.

Asian youths with ADHD and the older Asian adults were more likely to receive emergency services than their white counterparts. Fear of being stigmatized may delay family members from seeking care until problems are highly exacerbated, leading to use of emergency services. In addition, a lack of bilingual, culturally relevant treatment may impede retention in programs, as has been cited in multiple studies (23–25). Older Asian adults may receive few services because of cultural views held by less acculturated elderly that dissuade them from long-term use of psychiatric medications. Also, some may experience transportation difficulties because of isolated living arrangements.

Diagnoses

Psychoses were more prevalent among younger and older black adults than among their white counterparts, a finding consistent with studies that have identified the greater likelihood that clinicians differentially diagnose blacks as having psychoses rather than affective disorders (26). A recent study (27) based on race-blinded assessments sustained this finding, citing as a major reason that established diagnostic criteria may be insensitive to cultural variation in symptom expression. Other reasons cited in the literature include clinician failure to recognize the distinct cultural presentation of symptoms of blacks and to obtain sufficient information for culturally accurate diagnoses (28,29). Black youths have higher rates of disruptive behaviors than whites, and the possibility of clinical misdiagnosis has also been noted (30). Black youths may therefore receive inappropriate interventions, leading to their observed lower engagement rates.

Hispanic adults with anxiety disorders received services at substantially lower rates than their white counterparts. This finding is in contrast to those with depression, who received services at essentially the same rates as whites. This suggests that cultural presentation of symptoms may cause clinicians to miss depressive symptoms of persons with an anxiety disorder and to fail to treat these symptoms (31).

Asian adults ≤65 who sought treatment were more likely to have a psychotic disorder than another diagnosis, perhaps because those with depression made greater use of primary care or private mental health providers, whereas those with a psychotic disorder, because of its severity, required mental health services. Older Asian and Hispanic service users were substantially more likely than whites to have diagnoses of depression. Isolation from younger family members because extended-family structures are no longer commonplace could contribute to situational depression for both of these groups.

Recommendations for cultural competency activities

Screening in non–mental health settings.

To promote access, routine screening for mental disorders in primary care and for elderly adults in community locations might increase rates of accessing mental health services, particularly for Asians. Health care reform initiatives that integrate care across health sectors could improve this situation by ensuring routine screening in such settings.

Integration of services for black youths.

Service use rates of black youths with disruptive disorders were notably low. They are at high risk for community-based problems, which may impede their appropriate use of mental health services. Cross-system coordination of services and supports is addressed nationally in federally sponsored systems of care initiatives (32). For example, New York offers the Children’s Plan (33). Special approaches, however, may still be required to engage black family members into the care process and help them navigate and coordinate other youth services and support systems.

Culturally tailored mental health information.

Lower service use rates among older Hispanics and Asians may be due to their limited health literacy. The provision of culturally tailored mental health information suitable for them and their family members, in accessible languages and formats, that outline reasons for adhering to medications and the desirability of completing programs (34) could act to increase retention. Community-based mental health information campaigns that focus on specific cultural groups and that target social marketing (35) to community members may help to reduce the stigma of mental illness still strongly felt by the elderly members.

Assistance with transportation.

Providing assistance with transporation is likely to increase engagement of older adults who are isolated and may be of particular value among the less acculturated, who may have limited abilities to make arrangements and tap local resources (36, 37).

Cultural group–specific diagnostic training materials.

The higher prevalence of psychoses among black adults and of disruptive disorders among black youths and the lower service use rates for anxiety disorders in comparison with depression among Hispanics suggest that the distinct presentation of illness symptoms by members of a cultural group may not yet be fully understood. Specialized training materials for clinicians that address cultural presentation of symptoms and their implications for diagnosis and treatment (38,39) are required.

Cultural competency assessments.

Globally, differences in service use rates may diminish as programs become more linguistically and culturally competent. Assessments of organizational and program-level cultural competencies using instruments that identify areas that need improvement (40) are recommended to become part of mandated performance measurement of health care reform providers.

Limitations

To better represent access, arguably the denominator of treated prevalence should include only the number of persons in need rather than the total population. Although rates would have been higher with a smaller denominator, the magnitude and directionality of differences would be the same provided the proportions with mental illness are equivalent across groups, a notion supported by the literature (1). Some have suggested that private-sector services are of higher quality than state-regulated services. Therefore, although access to state-regulated services was greater for Hispanics and blacks, they may be receiving lower-quality services. To verify such speculation requires a special study. Notably however, many youths and adults with private insurance in New York choose to use the public mental health system. Annual weeks in service may not capture retention. We interpreted lower numbers of treatment intensity as reflecting an inability of the programs to retain clients to complete desirable treatment regimens, but other explanations are possible. Multivariate models, although desirable, could not be constructed because the procedures used to obtain annual estimates from the one-week sample moved the unit of analysis from the individual to an aggregated number in a group category (the number of blacks with psychoses, for example). The influential covariates of race-ethnicity, age, and diagnosis were accounted for through stratified analyses. Finally, according to the definition of the Institute of Medicine (41), differences in the study measures may not necessarily signal mental health care disparities. The requirement for controlling for clinical need was met by diagnostic-specific stratified analyses, but treatment preferences that might well affect utilization for some cultural groups are unknown.

New York State has a history of innovation in mental health service delivery and the promotion of cultural competency in its organizations and programs (42,43). Compared with other states, New York’s per capita expenditures for mental health services have been consistently high (44). Hispanics of Caribbean origin make up the largest majority of the state’s Hispanic population (58%), followed by those of South American origin (15%), Mexican origin (12%), and Central American origin (9%) (45). Against this backdrop, we saw that treatment prevalence rates for blacks and Hispanics were higher than those for whites and many instances in which racial-ethnic groups received services at rates equal to or higher than those for whites. The specific disparities that were noted have served to signal areas in which actionable steps can be taken. The granular level of analyses allowed suggestions to be made for cultural competency activities, some of which have already entered the public conscience, whereas others continue to require diligent attention.

The authors received support from the New York State Office of Mental Health to the Nathan S. Kline Institute Center of Excellence in Culturally Competent Mental Health.

The authors report no competing interests.