Spanish-Language Community-Based Mental Health Treatment Programs, Policy-Required Language-Assistance Programming, and Mental Health Treatment Access Among Spanish-Speaking Clients

Abstract

Objectives. We investigated the extent to which implementing language assistance programming through contracting with community-based organizations improved the accessibility of mental health care under Medi-Cal (California’s Medicaid program) for Spanish-speaking persons with limited English proficiency, and whether it reduced language-based treatment access disparities.

Methods. Using a time series nonequivalent control group design, we studied county-level penetration of language assistance programming over 10 years (1997–2006) for Spanish-speaking persons with limited English proficiency covered under Medi-Cal. We used linear regression with county fixed effects to control for ongoing trends and other influences.

Results. When county mental health plans contracted with community-based organizations, those implementing language assistance programming increased penetration rates of Spanish-language mental health services under Medi-Cal more than other plans (0.28 percentage points, a 25% increase on average; P < .05). However, the increase was insufficient to significantly reduce language-related disparities.

Conclusions. Mental health treatment programs operated by community-based organizations may have moderately improved access after implementing required language assistance programming, but the programming did not reduce entrenched disparities in the accessibility of mental health services.

Among the roughly 55.5 million persons in the United States speaking a non-English language at home in 2007, about 34.5 million spoke Spanish; of those Spanish speakers, more than 10 million spoke English “not well” or “not at all”¹ and were thus considered persons with limited English proficiency (LEP). Persons with LEP “are unable to communicate effectively in English because their primary language is not English and they have not developed fluency in the English language.”² In California, the state with the largest Spanish-speaking population in the United States, about 40% of persons aged 5 years or older among the state’s 14 million Latino/Hispanic population are considered persons with LEP.³

LEP intersects with sociocultural and immigration-related barriers, thus preventing mentally ill persons with LEP from receiving needed care. Being uninformed about mental illness and interpreting and expressing symptoms of mental illness as something other than mental illness by using a culturally preferred idiom of distress, as well as turning to family and community network members when seeking help who reinforce nonpsychiatric perspectives, can divert persons with LEP from the path to mental health specialty care.⁴ The stigma associated with mental illness,⁵ distrust of treatment bureaucracies (S. Leask and L. R. Snowden, unpublished data, 2012), and, for immigrants, fear of being challenged by authorities and asked to account for their immigration status⁶ create additional barriers.

Nonetheless, LEP introduces a significant barrier of its own. Persons with LEP find it difficult to communicate in English language–oriented health care settings,^7–9 and they often either do not receive needed health care or receive ineffective care.^10–20 Language proficiency may be especially challenging in mental health treatment because psychiatric evaluation hinges on obtaining a thorough history, and because key symptoms are not reflected in directly observable behaviors or signs of morbidity and can be elicited only via self-report.²¹ Language barriers can prevent recognizing and labeling mental health problems and can interfere with successful communication about treatment needs and care options.^22–26

According to legal interpretations, executive-branch directives, and US Office of Civil Rights enforcement actions, Title VI of the 1964 Civil Rights Act²⁷ requires recipients of federal funds to provide language assistance services to persons with LEP.^28,29 By 2008, every state had passed laws supplementing federal law, further requiring language assistance for persons with LEP seen in health care settings.³⁰

To assist in compliance, the US Office of Civil Rights issued guidelines for implementing Title VI requirements. These guidelines call for assessment of the language needs of service-eligible populations and development of written policies to meet needs; training of staff in language assistance policies and procedures; monitoring of the implementation of policy and procedures; provision of trained interpreters; translation of written materials; and notification of beneficiaries that they are entitled to translation assistance free of charge.³¹

Like other mental health agencies, in 1997 the California Department of Mental Health (DMH) adopted a “threshold language access policy” to meet its Title VI obligation for treating persons insured through the state’s Medicaid program, called Medi-Cal.³² Under the policy, threshold status is reached when either 3000 Medi-Cal enrollees in a county or 5% of the county’s Medi-Cal residents, whichever is greater, speak a non-English language. For threshold languages, the policy directs county mental health plans toward a 4-part response: (1) a 24-hour, toll-free phone line with linguistic capability; (2) translated written materials to assist beneficiaries in accessing medically necessary specialty mental health services, including personal correspondence; (3) bilingual clinicians or other bilingual nonstaff, or interpreters or telephonic translation capacity at intake appointments, assessment interviews, treatment sessions, and at other key points of contact; and (4) information to consumers and communities about the availability of these linguistic services, free of charge.

One previous study of the threshold policy’s impact in California found that the policy’s mandated language assistance programming increased access to mental health treatment for Vietnamese-speaking and Russian-speaking Medi-Cal beneficiaries, but it found no evidence that access increased for Spanish-speaking beneficiaries.³³ However, the study did not examine how counties implemented threshold language programming and could not detect differences associated with the mode of implementation.

Contracting with community-based organizations (CBOs) operating specialized treatment programs for non-English-language speakers^34,35 is an attractive option for implementing the required language assistance programming. CBOs are

not-for-profit organizations such as non-governmental, civil society organizations, or other grassroots organizations, overseen by an elected board of directors and guided by a strategic plan developed in consultation with community stakeholders.^36(p33)

They operate health and social programs, as well as LEP-focused mental health programs, to fulfill a wider community service mission.³⁶ CBOs seek strong community ties and pursue community oversight and governance; they also practice social, economic, and political advocacy, thereby promoting credibility and community trust.³⁶ Seeking the advantages enjoyed by CBO-operated programs, mental health officials sometimes establish specialized LEP-serving programs operating directly under their authority.

A handful of past reports indicated that specialized mental health programs for persons with LEP may be especially effective at bringing them into treatment. In 1 study, Latino and Asian persons with LEP received more outpatient care in such programs than their counterparts seen elsewhere, and their initial contact with a treatment program was significantly less likely to come about via emergency service encounters.³⁷ A second study found that mental health programs specializing in clients speaking Asian languages provided an alternative to threshold language policy requirements for bringing Asian-language speakers into treatment.³⁸

We investigated whether access rates for Spanish-language mental health treatment rose for persons with LEP when CBOs’ mental health treatment programs implemented the language assistance programming required by threshold language policy. We hypothesized that, because their goals are closely aligned with addressing the cultural and linguistic orientation and interests of Spanish-speaking communities, and because this enables them to reach out effectively to LEP community members, CBOs’ language assistance programs will promote greater treatment entry than programs that are directly county operated.

We also explored whether CBOs’ implementation of language assistance programming was effective and widespread enough to bring about a statewide reduction in the disparity in access between English and Spanish speakers.³⁹ We evaluated any potential increased access experienced by Spanish speakers within a larger framework of disparities in access to mental health treatment. To our knowledge, this study was the first covering a large and diverse region, including a substantial Spanish-speaking population, to assess the effect of CBOs’ implementation of language assistance programming on the accessibility of mental health services for Spanish speakers with LEP. We also assessed reductions in disparities in access to mental health treatment for Spanish versus English speakers.

METHODS

This study used a nonequivalent control group and a time series design, exploiting California’s decentralized Medicaid structure, in which county plans act independently. For counties implementing a program with Spanish as a threshold language, the onset of implementation was staggered over 39 annual quarters from 1997 to 2006. To assess the effect of language assistance programming, we identified 2 sets of counterfactuals. Counties implementing programming served as their own controls, before and after implementation. Other counties that did not implement language assistance programming for Spanish speakers also acted as nonequivalent controls. This group included counties with no threshold languages and counties not having Spanish as a threshold language but having threshold languages other than Spanish (e.g., Cantonese, Vietnamese, Russian) and implementing language assistance programming for languages other than Spanish.

Variables and Data Sources

Dependent variable: penetration rates by quarter and primary language.

We calculated penetration rates—the percentage of the beneficiaries receiving services relative to all beneficiaries eligible to receive services—for each county and quarter. Penetration rates have been widely used to measure treatment access.⁴⁰ Specifically, we calculated penetration rates according to the number of Spanish-speaking clients with LEP who received specialty mental health care in a given county and quarter (the numerator), divided by the number of LEP Spanish-speaking Medi-Cal beneficiaries who were eligible to use services in that county and quarter (the denominator). Both the numerator and denominator were restricted to LEP Spanish-speaking persons aged 19 to 64 years. To minimize extreme rate fluctuations affecting fractions with small denominators, fractions with small denominators—defined as counties with fewer than 50 enrolled Medi-Cal beneficiaries—were eliminated from further consideration.

To assess whether Spanish-language speakers used mental health services (the numerator), we used DMH Medi-Cal specialty mental health claims. DMH staff prepared files that included date of birth, gender, race/ethnicity, primary language, type of eligibility, county of responsibility, type of service, service dates, amount approved, service provider, and an encrypted client identifier. For each county and each quarter, we counted the number of clients whose primary language was designated as Spanish.

To measure the number of people who were enrolled in Medi-Cal and therefore eligible for mental health treatment (the denominator), we used a file provided by the California Department of Health Services for primary Spanish speakers containing beneficiary characteristics, including the program under which the person qualified for Medi-Cal and the county of responsibility in the first month of each quarter included in the study period. Using this data source, we estimated the unduplicated number of adult Medi-Cal beneficiaries by primary language, county, and quarter.

Independent variable: implementation of minimally adequate language assistance and Spanish-language programming.

To assess minimally adequate implementation and use of Spanish-language, community-based treatment programs, we sent e-mail surveys to all 57 county mental health departments. We developed our survey in collaboration with county ethnic service managers and the state chief of multicultural services, with the California Mental Health Directors Association’s Ethnic Services Committee serving as consultants and cosponsors. We sought to assess how, for adults, each county implemented each of the 4 requirements stipulated by the threshold language policy.

In each domain of inquiry and for each threshold language, we asked closed-ended questions specifying the type of language services that were implemented for each threshold language in each year from 1997 to 2006. We also asked supplemental open-ended questions. We e-mailed the self-administered survey in April 2008. Each survey took approximately 1 to 2 hours to complete, depending on how many threshold languages were in the county and the complexity of services provided.

As they prepared to complete the survey, respondents were encouraged to consult records and reports, contracts, and other supporting documentation, including compliance reports submitted to state officials. We sent reminder e-mails and called nonresponding counties at least twice before the survey’s closing date.

The survey asked respondents detailed questions about implementation of threshold language requirements at any time during the study period, for any languages surpassing threshold levels. We received completed surveys from 32 of the 48 counties with at least 1 language surpassing threshold levels (66.7% response rate).

We constructed a binary indicator to assess whether each county had implemented language assistance programming to a “minimally adequate” level.^16,17 Minimally adequate implementation of programming occurred when counties implemented each of the 4 services required by threshold language policy for Spanish, in each quarter, whether program implementation occurred during the fiscal year in which it was required.

Independent variable: Spanish-language community-based treatment programs.

The survey also asked about specialty non-English-language programs funded by each county between the years 1996 and 2008. In particular, the survey asked about the presence of county-operated and contractor-operated language-specific programs and the characteristics of these programs, including language groups served, the fiscal years in operation, and the types of services provided. Among 41 non-English specialty programs reported, 63% served Spanish speakers (M. M. Masland, C. Lou, and C. J. Peng, unpublished data, 2012). Of the counties eligible for this analysis (n = 34) that had a Spanish-language assistance program at any time during the analysis (n = 32), CBOs ran programs in 9.

Control variables.

We also controlled for county-level and county-plan-level characteristics potentially associated with surpassing threshold levels of Spanish speakers and increased penetration rates among Spanish speakers. To capture secular trends in penetration rates and related factors, we included (1) a quarterly time trend variable, (2) the number of eligible beneficiaries per 1000 residents, and (3) the penetration rate among eligible English speakers for each county quarter, calculated using the same method as used with Spanish speakers.³² Subsequently, we used penetration rates among English speakers to assess disparity reduction. We controlled for the presence of bilingual staff in county programs to ensure that penetration rates for Spanish speakers could not be explained by the characteristics of individual staff members rather than those of the programs themselves. We also included county per capita income, which allowed us to estimate the resources available for responding to the needs of non-English speakers seeking mental health treatment. Finally, we included the proportion of Republican voters to assess political receptivity to meet LEP language assistance needs, because in California’s polarized political environment, support for public spending on vulnerable populations reflects underlying attitudes toward mental health and toward persons with LEP.⁴¹

Analysis

We conducted linear multivariate regressions with fixed county effects to control for static intercounty differences, whether measured or unmeasured, and other variables entered as time-varying covariates.⁴² The unit of analysis was county × quarter Spanish-speaking penetration rate.

Specifically, we examined 3 models, each testing a different aspect of the relationship between the effect of language programming and the penetration rate among Spanish-speaking persons with LEP. We first analyzed a basic model, building on a previous study,³² but including an interaction term—“programming × contractor”—to evaluate whether the effect of language programming was greater when implemented by a CBO. In this context, the “use of contractor” estimate reflects the penetration rate increase shown in counties with CBOs that did not implement language-specific programming, relative to counties without a CBO or programming. In the second model, we included an indicator for the number of bilingual staff in programs, to confirm that its omission was not biasing results. Because of the skewed distribution of the variable, we used a log transformation in the analysis. Third, to detect the reduction of disparities in access between Spanish and English speakers, we included a second interaction term—“language programming × CBO × English penetration rate”—indicating whether the contractor-implemented programming changed the degree of convergence between rates.³⁹ Access disparities would be reduced if this 3-way interaction term was significantly greater than 1, indicating that the rate was growing faster for Spanish speakers than for English speakers.

RESULTS

In all periods included in this analysis, eligible English speakers experienced dramatically greater use of Medi-Cal mental health services than did eligible Spanish-speaking persons with LEP (Table 1). Furthermore, the discrepancy between the penetration rates for the 2 groups appeared to have increased over time, largely because of the growth in English speakers’ use of mental health services. Average penetration rates for Spanish-speaking persons with LEP were very low, remaining consistently around 1%.

TABLE 1—

Unadjusted Mean County-Level Penetration Rates of Mental Health Care Services, by Language Proficiency: California Medicaid Program (Medi-Cal), 1997–2006

Fiscal Year	English Speakers, %	Spanish Speakers, %
1996–1997	3.60	0.86
1997–1998	4.03	0.99
1998–1999	4.51	1.19
1999–2000	6.53	1.16
2000–2001	6.83	1.04
2001–2002	7.37	0.96
2002–2003	7.45	1.36
2003–2004	7.19	1.27
2004–2005	6.80	1.22
2005–2006	6.87	1.15

Note. Both English and Spanish figures use data from the 34 counties for which Spanish data were available. The percentages shown are mean figures.

When we compared the characteristics of counties using CBOs with those providing mental health services in-house, the 2 groups differed at statistically significant levels on all measured characteristics (Table 2). Counties using CBOs tended to be much larger, to be less Republican, to offer many more bilingual providers, and to be more likely to have implemented threshold language programming.

TABLE 2—

Characteristics of Medicaid Mental Health Programs and Agencies, by Mode of Care Provision: California, 1997–2006

Variable	County Run (n = 981)	CBO Run (n = 345)	P
Language access programming, %	67.48	76.23	.001
Threshold status,a %	61.26	70.43	.002
Penetration rate, English speakers, mean	5.82	6.90	< .001
Eligibility/1000, mean	602.84	4107.19	< .001
Per capita income, $1000, mean	27.17	33.92	< .001
Majority Republican voters, %	53.11	41.16	< .001
No. of bilingual providers, mean	17.26	180.41	< .001

Note. CBO = community-based organization. The unit of analysis is county quarter.

^aThreshold status is reached when either 3000 Medi-Cal enrollees in a county or 5% of the county’s Medi-Cal residents, whichever is greater, speak a non-English language.

After we controlled for county fixed effects and several time-varying factors, implementation of threshold language programming was not, by itself, associated with an increase in Spanish speakers’ penetration rates (marginal effect = 0.05 percentage points; P > .05), nor was the use of CBOs in offering mental health services, relative to county-offered services (–0.14 percentage points; P > .05; Table 3). However, when CBOs were employed while counties implemented language programming, as indicated by the programming × CBO interaction term, Spanish speakers experienced a moderate but statistically significant increase in their penetration rates (marginal effect  = 0.28 percentage points; P < .05), a 25% increase compared with the average penetration rate for Spanish speakers during this 10-year period of 1.12 percentage points, even after accounting for the number of available bilingual providers.

TABLE 3—

Factors Associated With Provision of Spanish-Language Mental Health Services for Persons With Limited English Proficiency (LEP): California Medicaid Program (Medi-Cal), 1997–2006

	Model
Variable	Main Model, b (SE)	Includes Bilingual Providers, b (SE)	Includes Disparity Reduction, b (SE)
Quarterly time trend	0.007* (0.004)	0.007 (0.004)	0.009* (0.004)
LEP programming	0.053 (0.075)	0.050 (0.075)	0.085 (0.077)
Programming × quarter	−0.002 (0.003)	−0.002 (0.003)	−0.003 (0.003)
Use of CBO	−0.137 (0.246)	−0.129 (0.247)	−0.082 (0.249)
Programming × CBO	0.280* (0.115)	0.277* (0.115)	0.027 (0.193)
English penetration rate	0.090** (0.008)	0.090** (0.008)	0.084** (0.009)
Eligible (1000)	0.000 (0.001)	0.000 (0.001)	−0.000 (0.001)
Per capita income (1000)	−0.034** (0.009)	−0.035** (0.009)	−0.039** (0.010)
Votes Republican	0.033 (0.080)	0.038 (0.080)	0.023 (0.080)
Bilingual providers, log	…	0.012 (0.015)	…
Programming × contractor × English penetration rate	…	…	0.034 (0.021)
Constant	1.355** (0.235)	1.362** (0.236)	1.478** (0.247)
Adjusted r2	0.093	0.093	0.094

Note. CBO = community-based organization. All models were run as linear regressions with county-level fixed effects. The number of counties in the study was 34, the number of yearly quarters was 39, and the number of counties × quarters was 1326.

*P < .05; **P < .01.

Additionally, the implementation of language programming by CBOs instead of by counties did not reduce access disparities between English speakers and Spanish speakers with LEP, since the estimate for the 3-way interaction was not significantly greater than zero (0.034 percentage points; P < .05).

DISCUSSION

Observing Medi-Cal Spanish speakers’ mental health penetration rates for nearly 10 years, we found that the threshold language policy’s language assistance programming package increased the penetration rates for Spanish-language persons with LEP, but only when the language assistance programming was implemented by CBOs. Both programs directly operated by counties and CBO-operated programs are widespread in California. For purposes of implementing threshold language–required language assistance programming, however, we found that access increased only for CBO-operated programs.

In CBO-operated programs, penetration rates rose by nearly 0.3 percentage points, a large increase considering that average penetration rates hovered around 1% of eligible Medi-Cal beneficiaries. Future research should identify access-promoting characteristics differentiating programs operated by CBOs from those operated by counties and other public providers of mental health services.

Having a greater bilingual staff presence—an oft-promoted remedy for treatment disparities—was not a significant consideration for promoting greater treatment access. Conceivably, it is through organizational ties and outreach-promoting practices more than a bilingual individual’s personal actions that improved treatment access comes about.

One perspective for viewing the mental health programming offered by a CBO is that of “organizational cultural competence,” a quality that may characterize successful LEP and ethnic minority–serving programs.⁴³ According to one cultural competence model developed after synthesis of a vast literature on the subject, culturally competent programs embrace a community-serving vision and mission, encourage communication with complementary community-serving programs and with the community itself, and promote community participation in program governance.⁴⁴ If CBOs act as culturally competent organizations, our findings are consistent with the expectation that culturally competent organizations are effective vehicles for promoting access to mental health treatment for persons with LEP.

On the other hand, improved treatment access for LEP Spanish speakers did not translate into wide-scale reduction of disparities in treatment access for English vs Spanish speakers. This failure can be explained by the fact that, although access rates increased for Spanish speakers, they increased even more for English speakers.

If Spanish speakers needed mental health treatment less than English speakers, the access differential we documented might be acceptable; all else being equal, less needy populations require less treatment.⁴⁵ In fact, LEP Spanish speakers appear to have lower levels of treatment need than do English speakers. Rigorously derived estimates indicate that rates of mental illness among non-Latino Whites—almost all English speakers—are about 1.8 times greater than rates for immigrant Latinos, about half of whom have LEP.⁴⁶ Yet in our study, English speakers’ access rates were more than 5 times greater, on average, than those of LEP Spanish speakers. This access differential suggests that a disparity exists above and beyond differences in treatment need between English and Spanish speakers.

This analysis has several limitations. First, it was confined to Medicaid-enrolled persons in California; whether the study’s findings generalize elsewhere is uncertain. Although we controlled for county fixed effects and several time-varying controls, we cannot rule out the possibility that Spanish-language penetration rates increased for unobserved reasons that were also correlated with language assistance programming.

In California, specialty language assistance programs constitute a key, largely overlooked component of the mental health treatment system for serving the many members of ethnic minorities who have LEP. More research is needed that builds on the sparse knowledge currently available. Future studies of specialty non-English-language programs should differentiate clearly between programs that are directly operated by public mental health officials and programs that are CBO operated, measure distinguishing organizational characteristics, and assess their service delivery consequences, both for treatment access and treatment effectiveness. The research agenda should include investigating how successful programs improve treatment access—whether it comes from advertising by a credible source, through outreach efforts by community health workers, or by capitalizing on established networks of community leaders and respected organizations.

Furthermore, greater emphasis must be placed on studying the quality of care that is provided. Greater treatment access is of little consequence if, whether for language-related or other reasons, ineffective or otherwise inadequate treatment is a result. We must seek a comprehensive, deeper understanding of treatment programs and care provided to the large LEP Spanish-speaking population.