Variation in Mental Health Service Use Among U.S. Latinos by Place of Origin and Service Provider Type

Disparate access to and utilization of mental health care is a long-standing concern for racial-ethnic minority populations (1–4). Previous studies identified multiple factors associated with this disparate utilization, including cost and lack of insurance (5,6), linguistic challenges (5–7), perception of mental health symptoms as general medical conditions (8,9), and lack of knowledge about where to access help (6,10). This disparity in service use is especially concerning in light of the rapid growth of the Latino population—now the largest racial-ethnic minority group in the United States (11). The Latino population is predicted to triple from 2005 to 2050 and will constitute 29% of the U.S. national population by 2050 (12).

Although a growing body of research has examined factors associated with mental health service utilization among Latinos, much of the literature has focused on Latinos collectively, without fully recognizing the diversity within the Latino population. A key component of this diversity is place of origin because of its association with unique migration histories and sociodemographic characteristics. Migration from both Mexico and Puerto Rico, for example, is traditionally associated with seeking higher wages, whereas Cuban migration has been driven by the socialist leadership. Notable rates of Mexican migration began with the establishment of two U.S. guest worker initiatives (bracero programs) in 1917 (13) and in 1942; the latter facilitated the migration of 4.2 million Mexicans for agricultural labor (14). Continued Mexican migration has largely centered on the search for higher-paying jobs than can be found at home.

Puerto Ricans were granted U.S. citizenship in 1917, which initiated some degree of migration; however, more substantial waves occurred after 1945 as a result of economic shifts in Puerto Rico (15). For example, about 470,000 Puerto Ricans moved to the United States after an industrialization program (Operation Bootstrap) that was launched in 1947. In subsequent decades, migration rates have fluctuated on the basis of economic circumstances at home, as well as the availability of employment in the United States (15). Cuban migration to the United States rose sharply beginning with the Cuban revolution in 1959, particularly as wealthier individuals sought refuge from the socialist regime (16). Rates of migration increased after the Cuban Migration Agreement in 1995 (17). As a population seeking asylum, Cubans can apply for access to humanitarian provisions not provided to other Latino migrants, which serves as a driving force for some to move to the United States.

In regard to sociodemographic characteristics of Latino subgroups, Cubans constitute relatively older and more educated individuals (18–21), and Puerto Ricans are more likely to be proficient in English (18–20). Cubans also report a higher level of cohesion among family members, compared with Mexicans and Puerto Ricans, who reported the lowest level (21).

In terms of mental health issues, Alegría and colleagues (22) found that foreign-born Latinos had lower rates of psychiatric disorders than their U.S.-born counterparts. However, when the Latino sample was disaggregated into various subgroups, the disparity remained only for Mexicans. This diversity accentuates the importance of understanding variations in the mental health of Latinos and their service utilization patterns. Although previous studies suggest the importance of understanding disparities in mental health service use by place of origin (22–27), limited research has examined the role of Latino subgroups in mental health service utilization. The exception is a recent study by Keyes and colleagues (27). However, their study was limited by the aggregation of various service types, and it did not account for variation in use by service provider type. Specifically, considering the potential barriers to mental health service use among immigrant populations (for example, unfamiliarity with the U.S. mental health system and limited English proficiency), understanding which types of professionals are utilized for services is critical. Thus our study extends the work of Keyes and colleagues by examining variation on the basis of provider type, including specialty mental health providers, general medical providers, and other types of providers.

Andersen’s (28) behavioral model of health service utilization was used as the framework for this study. The model conceptualizes health service use as a function of predisposition to use services, enabling conditions to secure services, and need for such services, and it has been widely used for various vulnerable populations, including racial-ethnic minority groups and immigrants because of its comprehensive approach to explaining their use of general medical and mental health services (29–36). We examined variations in the main effects of predisposing, enabling, need, and immigration-related factors and their interaction effects with place of origin on various types of mental health service use among nationally representative samples of Latino subgroups in the United States. By exploring variations in service use by place of origin and provider type, the results can help inform mental health policies and services for various Latino subgroups in the United States.

Methods

Data Source and Study Sample

Data were obtained from the National Latino and Asian American Study (NLAAS). The NLAAS is a nationally representative survey that estimates the prevalence of mental disorders and rates of mental health service utilization by Latino and Asian Americans in the United States (37). The NLAAS data were collected in 2002 and 2003 by in-person household interviews or telephone interviews in English or in the respondent’s native language; the weighted response rate was 75.5% for Latinos and 65.6% for Asians (38). Although this data set may be dated, the NLAAS is still one of the most comprehensive and largest studies of nationally representative Latinos and Asian Americans (39).

The study sample consisted of 2,533 Latino adults (age 18 and older). To make a clear comparison of Latino respondents on the basis of their place of origin, non-Latino respondents were excluded from this study. Respondents were classified into four subgroups on the basis of self-reported place of origin: Cuban, Puerto Rican, Mexican, and other Latino.

Measures

Mental health service use was the dependent variable. Respondents were asked whether they received services in the past 12 months from each of three provider types for problems with emotions, nerves, or use of alcohol or drugs. The three service provider types included specialty mental health services provided by psychiatrists, psychologists, and other mental health professionals and by mental health hotlines; general medical services provided by general practitioners, family doctors, nurses, occupational therapists, and other health professionals; and other services provided by social workers, counselors, religious or spiritual advisors, self-help groups, and Internet support groups (40,41). On the basis of use of the three types of providers (yes or no), we created another dichotomous variable to indicate whether the respondent used any type of mental health services (yes or no). This operationalization is a conventional way of measuring various types of mental health services in previous studies (40,41). However, it is important to note that this self-reported service use might be underestimated (37).

Predisposing factors included age (in years), gender, marital status, and educational attainment. Marital status was dichotomized: currently married or cohabiting versus divorced, separated, widowed, or never married. Educational attainment consisted of four categories: less than high school graduate, high school graduate, some university, and university graduate or above.

Enabling factors included poverty status, insurance coverage, and social network characteristics. Poverty status was based on the 2001 U.S. Census income-to-needs ratio (yes or no). We dichotomized insurance coverage to indicate whether the respondent had any type of insurance (insured versus uninsured). Two social network variables included frequency of contact with relatives and with friends who don’t live with respondents; respondents were asked how often they talked on the phone or got together (most every day, a few times a week, a few times a month, once a month, and less than once a month). To achieve parsimony, we used social network variables as continuous variables in multivariate models, with higher values indicating more contacts.

Psychiatric disorders, a need factor, were assessed by a modified version of the World Mental Health Composite International Diagnostic Interview, a comprehensive, fully structured psychiatric diagnostic interview developed by the World Health Organization to provide valid information about the prevalence of mental disorders in the general population (42). We created four dichotomous variables (yes or no) to indicate whether the respondent had the following disorders during the 12-month period: mood disorders (major depressive disorder, major depressive episode, and dysthymia), anxiety disorders (panic attack, panic disorder, agoraphobia without panic disorder, agoraphobia with panic disorder, social phobia, generalized anxiety disorder, and posttraumatic stress disorder), substance use disorders (alcohol abuse, alcohol dependence, drug abuse, and drug dependence), and eating disorders (anorexia, binge eating, and bulimia). We created an additional dichotomous variable (yes or no) to indicate whether the respondent had any of these four types of psychiatric disorders (40,41). It should be noted that bipolar disorder and schizophrenia were not measured in the NLAAS (43).

Immigration-related factors included length of stay in the United States (0–5, 6–10, 11–20, or ≥21 years), age at immigration (0–12, 13–17, 18–34, or ≥35 years), U.S. citizenship status (U.S.-born or naturalized citizen versus noncitizen), and English proficiency (excellent-good versus fair-poor).

Data Analysis

We used the survey procedures of SAS, version 9.3, to account for the NLAAS’s complex sampling design. To compare sample characteristics by place of origin, we conducted Rao-Scott chi square tests for all categorical variables (44) and calculated F statistics for a continuous variable of age. We conducted a series of logistic regression analyses to examine the factors associated with use of services from each provider type. To examine variation by place of origin (the moderating effect of place of origin on the relationships among covariates and mental health service use), we also tested potential interactions. By comparing Wald chi square statistics of two models (one with and one without interactions), an omnibus test was conducted (45). On the basis of a conservative cutoff point (variance inflation factor >4.0) (46), we found no multicollinearity issues in the models. Because of the relatively small observed missing values, listwise deletion was conducted.

Results

The mean±SE age of the sample was 37.98±.55, and mean age differed significantly by place of origin (F=37.8, df=3 and 56, p<.001). Results of Rao-Scott chi square tests showed that all predisposing, enabling, need, and immigration-related characteristics were significantly different (p<.05) by place of origin, with the exceptions of depression and substance use disorders (Table 1). Over one-fifth of respondents had any psychiatric disorders (21.3%), whereas only 9.6% reported that they received any mental health services in the past 12 months. Receipt of mental health services from specialty providers and other providers differed significantly (p<.05) by place of origin, whereas no variation was found in receipt of services from general medical providers.

Table 2 and Table 3 present the results of the logistic regression models. To examine the gross effect of place of origin, model A included only ethnic dummy variables; the results indicated that Puerto Ricans were more likely than Cubans, Mexicans, and other Latinos to use services provided by any mental health providers; however, no variation by place of origin was found in use of specialty mental health care. The results of omnibus tests indicated that the model specifying the interaction effects of place of origin (model C) had a better fit (p<.001) than the model not specifying these interactions (model B). Consequently, the following results are based on model C.

Regardless of service provider type, the presence of a psychiatric disorder was found to be a strong predictor of mental health service use (p<.001). Having insurance was also found to be a significant predictor of use, with the exception of services provided by others (social workers, religious advisors, and so forth). We found a significant interaction between Cuban origin and gender (odds ratio [OR]=4.26, p<.05) indicating that Cuban women were more likely than Puerto Rican women to receive services from specialty mental health providers. Similarly, Mexicans with a psychiatric disorder were more likely than Puerto Ricans with a psychiatric disorder to receive mental health services from general medical care providers (OR=3.23, p<.05). We also found a significant interaction between Cuban origin and age (OR=.96, p<.05) indicating that the effect of age on use of services from other providers was smaller among Cubans than among Puerto Ricans. The likelihood of using services from other providers was higher among those with better English proficiency (OR=2.43, p<.05). However, more frequent contact with relatives decreased the probability of using specialty mental health care (OR=.80, p<.05).

Discussion

The results provide important information on mental health service utilization among Latino subgroups, particularly in three provider categories: specialty mental health providers, general medical providers, and other providers (social workers, counselors, religious or spiritual advisors, self-help groups, and Internet support groups). Furthermore, we examined the effect of place of origin on service use to account for the heterogeneity among Latino subgroups. Using a nationally representative sample of Latinos, our study contributes to the literature by expanding knowledge of mental health service use by place of origin and service provider type.

The rate of service use was low (9.6%) in light of the percentage of respondents who had a psychiatric disorder (21.3%). This discrepancy has been noted in prior research, which has indicated that Latinos experience substantial access barriers and utilize mental health treatment at lower rates than non-Hispanic whites (1,24,47–51). As in other research on racial-ethnic minority populations (52), respondents who sought mental health services were more likely to visit general medical providers and other providers than to visit mental health specialists. This finding held even among individuals with health insurance, which may reflect the vulnerability of uninsured Latinos in terms of mental health service use (5). Notably, previous research found that Latinos were less likely than non-Hispanic whites to be insured (53). In our study, approximately 35% of the sample was uninsured.

Findings also indicated that Puerto Ricans were more likely than Cubans and Mexicans to use services, regardless of provider type. Considering the high rate of psychiatric disorders among Puerto Ricans (29.9%), which is consistent with previous findings (25,26,43), this result is not surprising. The presence of a psychiatric condition (a need factor) facilitates service use according to the Andersen model. However, to better understand the need for services, further studies should account for the level of functioning and severity of psychiatric symptoms. In addition, among Puerto Ricans, high rates of insurance (83.9%; enabling factor), citizenship (98.9%; immigration-related factor), and proficiency in English (69.2%; immigration-related factor) may facilitate mental health service use. By adding immigration-related factors to the original Andersen model, our study supported the generalizability of the extended Andersen model for racial-ethnic minority groups (41,54).

Findings also indicated the existence of variation in mental health service use by provider type among Latino subgroups. The very low rates of specialty mental health service use are particularly concerning, as is the higher likelihood among Mexican respondents of accessing mental health care from general medical providers. Minority populations are often treated for mental health conditions by general practitioners instead of specialty mental health providers (52,55). Although this may have benefits, such as convenient access, it may also yield poorer outcomes because of no treatment or inadequate treatment (2,56,57). Further, Lawrie and colleagues (58) demonstrated that practitioners in general medical settings were unwilling to include people with psychiatric disorders on their patient list. These providers’ attitudes toward and inadequate treatment of individuals with psychiatric disorders may influence the quality of and satisfaction with mental health services provided. Thus use of specialty providers is critical to adequate treatment provision.

Also, because individuals with psychiatric disorders tend to have comorbid general medical conditions (59,60), ensuring the integration of general medical and behavioral health care (61) could facilitate appropriate and adequate mental health services for this vulnerable population. Integrated health settings, such as medical homes with mental health services or behavioral health homes with general medical services, may improve treatment utilization and outcomes. Further research should address the quality of services and cost savings by comparing integrated models to better enhance access to effective care for this vulnerable population, which often seeks mental health care in primary care settings.

Identifying where Latino subgroups seek mental health services helps providers be fully prepared to assess patients’ psychiatric needs. Because Mexicans were found to be more likely to visit general practitioners, these providers should be aware of the need to assess and fully address (via sufficient treatment or referral) mental health needs. Policies that support integrated health care, such as the Affordable Care Act, help promote comprehensive and effective treatments. However, service provision must also address barriers unique to populations from each place of origin, recognizing how cultural norms, understanding of mental health, citizenship status, insurance barriers, and other factors influence service utilization. For example, given that Mexicans sought mental health treatment from general practitioners more often than did other Latino subgroups, they may benefit from direct written and oral education (in Spanish) on mental health diagnosis and treatment in the general medical setting. Further, general practitioners should link Mexican patients to affordable, culturally and linguistically appropriate mental health specialty services by building connections with Spanish-speaking providers in their communities who offer sliding scale—or other affordable—options for mental health treatment. Finally, providers with substantial knowledge of the cultural norms and immigration histories and patterns of each subgroup should tailor health assessment and education to the distinct experiences of patients. A further understanding of how the settlement process by subgroup contributes to health needs will facilitate appropriate diagnosis and treatment.

Some study limitations should be addressed. First, self-reported service use might be underestimated. Second, the study could not control for the severity of psychiatric symptoms, which play an important indicator of need in mental health service use, and for specific immigration status, such as legal and refugee status, which may better explain variation in mental health service use.

Conclusions

This is the first study to examine variation in mental health service use among Latino subgroups by type of service provider. Findings suggest the existence of differences among Latino subgroups in patterns of use and predictors of use. By highlighting differences in service use among Cuban, Mexican, and Puerto Rican subgroups, our findings also inform the provision of effective outreach and services to a population that underutilizes mental health treatment. Further studies should examine variations among Latino Americans in service use by immigrant legal status.