Psychometric Properties of Three Measures of Stigma among Hispanics with Depression

Abstract

While many measures of mental illness stigma have been developed, few have been validated in Hispanic populations. This study examined the psychometric properties of three stigma measures (Stigma Concerns about Mental Health Care [SCMHC], Social Distance Scale [SDS], and Latino Scale for Antidepressant Stigma [LSAS]) among a depressed, Hispanic sample. Data were collected during baseline assessments for two studies taking place in primary care settings (N = 500). Psychometric and factor validity were tested for each measure. Confirmatory factor analyses indicated adequate model fit, and adequate internal consistency reliability was found for all three measures. Stigma scores significantly differed by education level and gender. Findings from this analysis provide support for the use of the SCMHC, SDS, and LSAS in a depressed, Hispanic population. Assessing barriers to depression treatment, including stigma, are critical in engaging Hispanics in care and eliminating disparities for the population.

Introduction

Hispanics account for approximately 18.5% of the United States population (1), making them the largest minority group in the country, and are projected to make up more than a quarter of the nation’s total population by 2060 (2). While Hispanics demonstrate similar rates of psychiatric disorders compared to those of their non-Hispanic, White counterparts (3), Hispanics are less likely to utilize mental health services, including the use of prescription medications, and are more likely to discontinue treatment prematurely (4-7). These disparities in mental health treatment result from a complex set of sociodemographic, structural and cultural factors including lack of insurance, geographic inaccessibility, limited number of mental health specialists, language barriers, low mental health literacy, and stigma (6, 8, 9).

Stigma is a significant barrier to mental health care that disproportionately impacts mental health help-seeking of certain groups, including ethnic minorities (10). Cultural values of certain ethnic minority groups, such as Hispanics, potentially contribute to the stigmatization of people with mental illness and the use of mental health services (11). Compared to non-Hispanic whites, Hispanics report greater shame or embarrassment related to having a mental illness (12). Hispanics are also more likely to endorse negative attitudes towards antidepressant medication (i.e. the medication is not helpful and/or addictive), which are strongly related to causal beliefs about depression that involve interpersonal and situational factors rather than biological, genetic, or chemical factors (13, 14). Among Hispanics, stigma is negatively associated with the desire to engage in mental health care, management of depression symptoms, disclosure of mental illness to family and friends, and adherence to medications (15-17).

Stigma towards mental illness is a complex construct that lacks consistency and clarity in its conceptualization by researchers, which has led to the development of a wide range of measurement tools (18). In their review of mental illness stigma measures developed between 2004 and 2014, Fox and colleagues (18) identified more than 400 measures assessing mental illness stigma and categorized them as measuring different ‘stigma mechanisms’ based on whether the measure targeted stigma in those with mental illness and those without mental illness. However, the authors excluded measures of stigma towards mental health treatments from their review. While sometimes used synonymously, mental illness stigma and treatment stigma are distinct constructs that uniquely predict help-seeking outcomes (19, 20). For example, Tucker and colleagues (20) found that self-stigma of seeking mental health treatment was significantly related to self-blame while self-stigma of having a mental illness was not, suggesting that the behavior of seeking mental health treatment is viewed as more blameworthy since it represents an active decision to engage in a set behavior, whereas having mental illness may be seen as less of a “choice.”

Adding an additional level of complexity to stigma measurement is its measurement across different ethnic/cultural groups. Cultural factors may be key to understanding how stigmatization is shaped among different groups (21, 22); however, the assessment of stigma that incorporates cultural elements is lacking. In a review of studies assessing the use of mental illness stigma measures among non-Western European cultural groups, researchers found that most studies utilized adaptations of existing Western-developed stigma measures that were not designed to assess culture-specific forms of stigma (23). While the use of existing measures may be preferable to the development of new measures (18), steps should be taken to assure that measures are linguistic and culturally relevant for use with new populations (24). Previous research on stigma measurement among Hispanics has emphasized the need for this type of validation, with one study reporting different factor structures for a measure tested in both a Hispanic and non-Hispanic, White sample (25) and another study concluding a lack of support for a previously validated measure when it was tested with a primarily Spanish-speaking sample (8).

Research on psychometric properties of mental illness stigma measures among Hispanics is scarce, with few measures undergoing systematic psychometric validation (26). In 2010, Interian and colleagues (8) found evidence for the use of three different stigma measures with a sample of predominantly Spanish-speaking, Hispanic primary care patients (N = 200) who screened positive for depression. The three measures studied included adaptations of two previously established measures: Stigma Concerns about Mental Health Care (SCMHC), an adaptation of a measure used to assess broader barriers to depression treatment utilization (16), and the Social Distance Scale (SDS), a measure of discrimination used to assess people’s willingness to engage in social contact with people from other groups (27). Both measures were adapted to measure stigma specific to individuals with depression or a history of depression treatment. The third measure, the Latino Scale for Antidepressant Stigma (LSAS), was developed based on a qualitative study by Interian, et al. to measure stigma towards antidepressant medications specific to the Hispanic/Latino population. Both English and Spanish versions of the measures were used within the study. Spanish translations of the measures were developed by a bilingual research assistant and checked by two researchers, with modifications made until a consensus was reached. The measures are designed to display both the English and Spanish versions simultaneously.

In their study, Interian and colleagues (8) administered the measures at two time points five months apart and demonstrated support for the measures across the psychometric domains evaluated including factor analysis and internal consistencies. All three measures were found to be single-factor measures with internal consistencies ranging from .66 to .75. Additionally, the researchers assessed convergent validity of the measures by examining their correlations to one another and criterion-related validity by examining their relationship to three different utilization outcomes: whether participants were currently taking antidepressant medication, whether they had received any emotional care in the past three months, and whether they had ever received treatment for depression. The researchers found higher scores on the SCMHC to be significantly related to all three treatment utilization outcomes, whereas higher scores on the LSAS were associated with decreased odds of taking antidepressant medications. Results of the correlations revealed that the measures shared some common variance but assessed unique constructs related to stigma towards depression (8).

Since the development and testing of the SCMHC, SDS, and LSAS by Interian et al. (8), intervention studies have used versions of the measures to assess stigma towards depression and depression treatment. In one sample of predominantly Hispanic adults attending a community adult school, Unger and colleagues (28) utilized the SCMHC and an adapted version of the LSAS and reported internal consistencies for the measures to be .84 and .80, respectively. Hernandez and Organista (29) used the SCMHC and LSAS in their study with Spanish speaking immigrant Latinas at high risk for depression and similarly reported internal consistencies for the measures to be .83 and .79. Different versions of the SDS have been used to assess mental illness stigma among Hispanics with internal consistencies ranging between .58 for a 3-item version assessing stigma towards depression treatment (30) and .90 for a 5-item version assessing stigma towards suicide (31).

Despite their continued use over the last decade, no additional psychometric work has been performed on the Interian et al. (8) measures to provide support for their use in clinical settings or within intervention studies that specifically target stigma mechanisms among Hispanic patients. Thus, the purpose of the current study was to replicate the Interian study among a sample of Hispanic, primary care patients diagnosed with depression and seeking mental health care treatment. Specifically, we examined the measures’ internal consistency reliability, construct validity through confirmatory factor analyses, and convergent validity by examining the correlations to one another. As an additional analysis of construct validity, we examined the measures’ relationships to depression and anxiety severity among participants as well as demographic characteristics of age, gender, marital status, education level, and language. While the Interian et al. (8) study examined correlations between depression severity and stigma scores, we wanted to further this investigation by also exploring the correlations for anxiety severity, which is a disorder commonly comorbid with depression within primary care settings (32).

Methods

Overview

This study utilized baseline data from two studies on the integration of mental health services within a Federally Qualified Health Center: (1) Project DESEO: Depression Screening and Education: Options to Reduce Barriers to Treatment (ClinicalTrials.gov: NCT02491034) (33) and (2) METRIC: Measurement, Education and Tracking in Integrated Care: Strategies to Increase Patient Engagement and Reduce Mental Health Disparities among Hispanics (ClinicalTrials.gov: NCT02702596) (34). Both studies were reviewed and approved by the Institutional Review Board of the University of Texas at Arlington.

Sample

Purposive sampling was used to recruit participants for both studies who met the following inclusion criteria: 18 years or older, self-identified as Hispanic, met diagnostic criteria for depression, and were not currently receiving treatment for depression. Recruitment took place at a Federally Qualified Health Center’s three locations within a large metropolitan area in North Texas. Three-hundred fifty participants were enrolled for Project DESEO between February 2015 and October 2016 and 150 participants were enrolled for METRIC between February 2016 and February 2018. There was no overlap in patients in the two samples. All measures were collected at time of enrollment and were administered in either English or Spanish, based on patient preference.

Measures

Demographic information.

Demographic variables collected from participants via medical record extraction included their age, gender, primary language, marital status and education level.

Depression and anxiety.

Depression symptom severity was measured using the 9-item Patient Health Questionnaire (PHQ-9) (35), a self-report measure that assesses the frequency of the nine Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (36) depression symptoms within the last two weeks. Responses for each item range from 0 for “not at all” to 3 for “nearly every day,” with total possible scores ranging from 0 to 27. Scores greater than or equal to 10 are considered to indicate clinically significant depressive symptoms. Both the English and Spanish versions of the PHQ-9 have demonstrated strong internal consistency and similar factor structures in Hispanic samples (37, 38). A confirmatory factor analysis (CFA) and latent profile analysis were conducted with the PHQ9 in this sample demonstrating general support for the measure with Hispanics (39).

The Generalized Anxiety Disorder 7-item scale (GAD-7) (40) was used to measure anxiety symptom severity. Frequency of symptoms for GAD over the last two weeks are assessed based on the diagnostic criteria for generalized anxiety disorder in the DSM-IV, with responses ranging from 0 for “not at all” to 3 for “nearly every day.” Total possible scores range from 0 to 21, with scores of 10 or greater indicating probable cases of GAD. The GAD-7 has been found to be a reliable and valid measure for use with Hispanic Americans and has demonstrated strong internal consistency reliability for both the English and Spanish versions (41).

Stigma.

Three measures that were used by Interian et al. (8) were used to assess stigma towards depression and depression treatment. First, the Stigma Concerns about Mental Health Care (SCMHC) scale is a 3-item scale that assesses the anticipated stigma participants might feel if they were to seek treatment for depression by asking them if they agree or disagree to the three presented statements (example: “I would not want to receive treatment for depression because of being afraid of what others might think.”). Response categories for the statements include “0 – Disagree,” “1 - Agree,” and “7 – Don’t Know/Refuse.” Total scores for the scale are calculated by taking the sum of the three items while excluding the values of 7, with possible scores ranging from 0 to 3.

The Social Distance Scale (SDS) is a 6-item scale that measures social distance desirability from someone who has a history of depression treatment by asking questions such as “Would you be friends with someone who is or had been in treatment for depression?” and “Would you invite a person into your home who is or had been in treatment for depression?”. Response options include “0 – No,” “1 – Maybe,” “2 – Yes,” and “7 – Don’t Know/Refuse.” Total scores are calculated by taking the sum of the items while excluding any values of 7. Possible scores range from 0 to 12, with lower scores indicating greater desired social distance from someone who has or who is receiving depression treatment.

Finally, the Latino Scale for Antidepressant Stigma (LSAS) includes seven items measuring stigma towards the use of antidepressant medications. The seven items present stigma-related stereotypes pertaining to the use of antidepressants (example: “People who take prescription medicine for depression have a difficult time solving their problems on their own.”) and ask participants to identify what they believe other people think. Response categories include “0 – No one thinks that way,” “1 – Some people think that way,” “2 – Everyone thinks that way,” and “7 – Don’t Know or Refuse to answer.” Total scores are calculated by taking the sum of the items without including the values of 7. Possible scores for the scale range from 0 to 14, with higher scores indicating greater stigma.

Statistical Analyses

CFA modeling was used to assess the fit of the three stigma measurement models to the baseline DESEO and METRIC data. Responses of “7 – Don’t Know/Refuse” to any of the items on the scales were coded as missing. Due to low amounts of missing data in the SCMHC and SDS measures and coding of ”Don’t know” responses, weighted least squares means and variance adjusted (WLSMV) estimation in Mplus 8.2 was used to account for the ordinal nature of the data. WSLMV may produce less biased estimates of factor loadings given fewer response options per item when compared to other estimators for ordinal data such as maximum likelihood parameter estimates with robust standard errors, makes no distributional assumptions about the data, and has the added advantage of providing indices of model fit (42). Model fit to the data was indicated by a lower and non-significant χ² value, χ²/df value less than 3.0, and RMSEA scores less than .08. Additional fit indices included the Bentler Comparative Fit Index (CFI), Tucker-Lewis index (TLI), and weighted root-mean-square residual (SRMR) (46). CFI and TLI scores .95 or higher and a SRMR less than .10 indicate model fit with the data. Modifications to the model were included where correlation of error terms would generate a model χ² difference greater than 9.0 and supported conceptually by the measure. All factor loadings reported were standardized loadings. The SCMHC measure contained three items leaving the model as just identified with no available indices of model fit. Additional analyses included tests of reliability and psychometric validity and were performed in SPSS 25.0 and R stats psych package (47). These analyses examined associations between the stigma measure scores and patient demographic characteristics and reported depression and anxiety scores. These analyses included internal consistency reliability estimates (Cronbach alpha, α, and coefficient omega, ω) (48), Pearson’s r, and examining between groups differences on measure scores using ANOVAs and t-tests.

Results

Patient Characteristics

The sample of 500 Hispanics that met inclusion criteria for both studies (Table 1) included 460 women (92.0%) and nearly all Spanish-speaking (n = 468, 93.8%). Mean age of the sample was 38.98 years (SD = 10.16), and most were either married or cohabiting with a partner (n = 353, 71.0%). Educational attainment was relatively low with most of the sample report not completing high school or equivalent (n = 297, 59.4%). Overall mean PHQ-9 scores were 16.88 (SD = 4.09), and mean GAD-7 anxiety scores were 12.85 (SD = 4.63). Based on PHQ-9 scores, 354 participants (70.8%) reported either moderately severe (scores of 15 – 19) or severe (scores of 20 – 27) depression.

Table 1:

Descriptive Statistics of Sample

Demographic and Patient Characteristic	Total Sample (N = 500)
Age, M±SD	38.98 ± 10.16
Gender, female, n (%)	460 (92.0%)
Spanish Speaking, yes, n (%)	468 (93.8%)
Marital Status, n (%)
Married/cohabitating	353 (71.0%)
Never married	51 (10.3%)
Widowed	10 (2.0%)
Divorced	50 (10.1%)
Other	33 (6.6%)
Education Level, n (%)
8th grade or less	182 (37.2%)
Some high school	115 (23.5%)
High school or GED	127 (26.0%)
Vocational or trade school	10 (2.0%)
Some college	40 (8.2%)
College degree	15 3.1%)
Patient Health Questionnaire-9 (PHQ-9), M ± SD	16.88 ± 4.09
No depression (0-4), n (%)	0 (0%)
Mild depression (5-9), n (%)	10 (2.0%)
Moderate depression (10-14), n (%)	135 (27.1%)
Moderately severe depression (15-19), n (%)	215 (43.1%)
Severe depression (20-27), n (%)	139 (27.9%)
Generalized Anxiety Disorder scale (GAD-7), M ± SD	12.85 ± 4.63
Stigma Concerns About Mental Health Care, M ± SD	0.44 ± 0.84
Latino Scale for Antidepressant Stigma, M ± SD	6.11 ± 3.37
Social Distance Scale, M ± SD	9.47 ± 2.86

Confirmatory Factor Analyses

Across the SCMHC and SDS measures, there were 5% missing data per item or less. Data were assumed to be missing completely at random after statistically non-significant Little’s test (p > .05). For the LSAS measure, actual non-response ranged from 0.4% to 0.8%, however, 16.8% to 26.2% of respondents per item reported “don’t know”. This contributed to a much higher rate of missing data in the analyses with this measure.

SCMHC.

CFA for the SCMHC measure (Table 2) indicated good fit with higher loadings between .836 and .989 (all p<.001). The SCMHC measure only contained three items leaving the model just-identified. Therefore, no degrees of freedom remained to calculate other measures of model fit. Overall, the model explained between 69.9% and 97.9% of the variance in the three SCMHC items. Internal consistency reliability was acceptable with α = .749 and ω = .76.

Table 2:

CFA Factor Loadings and Model Fit Estimates for Stigma Measures

Measure	Items	Factor loadings, standardized	Item r2	Reliability		Fit indices
				α	ω	χ2 (df)	χ2 /df	p	RMSEA	90% CI	CFI	TLI	SRMR
SCMHC				.749	.76	_a	_a	_a	_a	_a	1.00	1.00	_a
	1	.836	.699
	2	.989	.979
	3	.864	.746
SDS				.754	.81	24.29 (9)	2.70	.004	.059	.031, .087	.991	.985	.032
	1	.737	.544
	2	.885	.802
	3	.888	.788
	4	.849	.720
	5	.759	.575
	6	.672	.452
LSAS				.846	.84	39.86(14)	2.85	.001	.062	.040, .085	.987	.980	.034
	1	.731	.534
	2	.752	.565
	3	.686	.471
	4	.768	.589
	5	.782	.612
	6	.761	.580
	7	.697	.485

^_aSCMHC measure contains three items having a CFA that is justified. Therefore, no degrees of freedom remained to calculate other measures of model fit.

SDS.

The SDS had excellent fit to the data (χ²=24.29, df=9, p=.004; χ²/df=2.70; RMSEA=.059; CFI=.991; TLI=.985; SRMR=.032). Items significantly loaded on the single latent variable with loadings between .672 and .895 (all p<.001). Item variance explained by model ranged from 45.2 to 80.2%. Internal consistency reliability was good with α = .754 and ω = .81.

LSAS.

Lastly, LSAS demonstrated similarly excellent fit (χ²=39.86, df=14, p<.001; χ²/df = 2.85; RMSEA=.062; CFI=.987; TLI=.980; SRMR=.034). Factor loadings were all significant (ranged from .686 to .782, p<.001). Between 47.1% and 61.2% of the variance was explained in the items. Internal consistency reliability was good with α = .846 and ω = .84.

Correlations among stigma measures.

Correlations (Table 3) are the estimates between the latent factor scores from a CFA where all three measures were allowed to correlate and were not strongly correlated despite being significant indicating each measure may assess a unique construct of stigma toward mental health and treatment. The SDS measure had a weak and negative correlation with SCMHC scores (r = −.225, p = .003), and considering lower SDS scores indicate the need for greater social distance, the need for greater social distance was associated with greater stigma towards mental health treatment (i.e., SCMHC scores). SCMHC and LSAS factor scores were weakly, positively correlated (r = .192, p = .006) indicating some association between stigmas concerning mental health treatment and psychotropic medication use.

Table 3:

Correlations between Stigma Measures

	SCMHC	SD	LSAS
SCMHC	-	−.225**	.192**
SD	−.225**	-	−.105
LSAS	. 192**	−.105	-
Age	−0.078	0.122	0.087
PHQ9 Depression	−0.075	0.011	0.164**
GAD7 Anxiety	−0.098	−0.039	0.161**

Note:

^*p < 0.05

^**p < 0.01

Stigma and Patient Characteristics

Stigma scores significantly differed by education level (Table 4) with less education associated with increased stigma towards mental health care (SCMHC, F [2, 483] = 4.13, p = .017) and social distance (SDS, F [2, 483] = 4.13, p = .017), and decreased stigma over use of medication (LSAS, F [2, 483] = 5.16, p = .006). Comparing across education groups, those with some college or more reported less social distance stigma and stigma towards mental health care than those of lower education. However, those with some college or more education reported significantly more stigma towards the use of antidepressant medications. Compared to men, women reported greater desired social distance from those with a history of depression treatment (t (494) = 2.50, p = .016).

Table 4:

Bivariate Analyses of Stigma Measures and Patient Characteristics

	n	SCMHC	SDS	LSAS
Gender		t = 1.183	t = 2.503*	t = 0.429
Male	39	0.59 (0.94)	10.37 (2.26)	6.33 (3.18)
Female	458	0.42 (0.83)	9.39 (2.89)	6.09 (3.39)
Marital Status		t = 1.135	t = 0.767	t = 0.399
Not married	144	0.36 (0.80)	9.62 (2.77)	6.22 (3.50)
Married	350	0.45 (0.84)	9.41 (2.88)	6.09 (3.32)
Educational attainment		F = 4.133*	F = 4.126*	F = 5.163**
Less than high school	295	0.51 (0.90)	9.25 (3.00)	5.76 (3.43)
High school	126	0.41 (0.82)	9.40 (2.68)	6.52 (3.24)
Some or more college	65	0.18 (0.53)	10.38 (2.47)	7.06 (3.12)
Spanish speaking		t = 1.004	t = 0.325	t = 1.404
Yes	465	0.45 (0.85)	9.46 (2.87)	6.06 (3.37)
No	31	0.29 (0.64)	9.63 (2.75)	6.94 (3.32)

Note:

^*p < 0.05

^**p < 0.01

Anxiety severity (r =.164, p = .007) and depression scores (r =.161, p = .003) were each weakly correlated with LSAS scores while neither were significantly correlated with scores the SMCHC and SD measures (Table 3).

Discussion

We found support for the use of the three Interian et al.(8) stigma measures - SCMHC, SDS, and LSAS - in a large sample of depressed Hispanic primary care patients, which adds to the developmental empirical research on stigma specific to primary care, low income, primarily Spanish-speaking patients (17). Similar to Interian et al. (8), we found evidence for the single-factor structure of all three measures as well as acceptable internal consistency reliability. While our sample shared many characteristics similar to those of the Interian study (primarily female and Spanish speaking with less than a high school education), they did differ in terms of the presence of clinically significant depression symptoms in the current study which adds meaningful support for a measure of stigma in a sample actively struggling with the disorder.

In our examination of the effects of clinical characteristics on stigma scores, we found increased depression and anxiety severity to be weakly correlated with stigma concerns towards antidepressant medication. These findings are consistent with Interian et al. (8) who also found a consistently weak correlation between depression severity and antidepressant stigma. Furthermore, Cheng et al. (49) found increased depression and anxiety scores were associated with increased self-stigma and perceived stigmatization by family and friends among Latino college students. In terms of demographic characteristics, females reported greater desire for social distance from people with depression. We also found stigma scores on all three measures to vary significantly by education level. In a more focused examination of this relationship in a subsample from the current study (50), we found that higher education levels were associated with decreased stigma towards seeking treatment for depression and desired social distance from those with depression, but an increased stigma towards antidepressant medication specifically. The unanticipated relationship between higher levels of education and greater stigma towards antidepressant medication among the participants in our current study warrant further investigation, especially given that low antidepressant use and non-adherence are common in Hispanic populations (4, 7).

Consistent with Interian et al. (8), our study also found primarily weak correlations between the measures and an insignificant relationship between the SDS and LSAS. These findings appear to confirm that the measures assess different domains of mental health-related stigma and understanding their unique contribution in measuring the stigmatization of mental health treatment, specifically, has important implications for predicting help-seeking outcomes. Primary care patients are often silent about their depression symptoms as a way to avoid stigma, which is a significant barrier to diagnosis and treatment (51). In our previous examination of mental health literacy, we found the more an individual knew about depression and its symptoms, the less likely they were to experience stigma about accessing mental health care services (50), however we also found that stigma specific to anti-depressant treatment increased with increased education about depression and depression treatment options (52). These findings suggest that efforts to reduce stigma toward mental illness and its treatments may require unique approaches and targets.

Increasing knowledge about mental health disorders may not be sufficient to address stigma towards treatments. Alternative strategies that normalize the act of seeking treatment, including discussions about the benefits and of myths of treatment, should be considered (19). Recent research suggests, however, even nuanced information about treatment approaches can affect beliefs about the causes and courses of different disorders. For example, O’Connor and Vaughan (53) found emphasizing psychological interventions for depression led to causal attributions of personal weakness, while endorsing medication treatment weakened confidence in individual control over the course of the illness. Such attributions may lead to unintended stigma and false self-blaming or, worse, a fatalistic attitude about the future course of the illness.

Even though effective and well tolerated treatments are widely available, most Americans with depression go untreated or undertreated. While the study sample was entirely Hispanic and their country of origin unknown, 83% of Hispanics in Texas are of Mexican descent (54) and there are significant disparities in receipt of guideline concordant care for Mexican Americans, which are only partially explained by lack of health insurance (55). Stigma is thought to be a major factor contributing to negative treatment outcomes and deficiencies in care and is associated with low antidepressant use, poor management of symptoms, and missed medical appointments (10), especially among Spanish-speaking Hispanic, (17, 56). Additionally, Vega, et al (17) found stigma among Hispanic primary care patients was associated with unwillingness to tell family or friends about their depression and lower likelihood of seeing a mental health professional, further compounding their depressive symptoms and social isolation.

In practice, it has been suggested that the stigma measures be transformed into a checklist or conversational Spanish to be used by clinicians to introduce depression in a culturally sensitive, effective manner with patients (17). Since Hispanics are less likely to visit a specialty mental health provider (57), primary care providers are in an optimal role to improve the quality of depression treatment through routine screening, use of sensitive probes to gather history, and a comprehensible explanation of depression and treatments (17). Fortunately, effective strategies for reducing stigma and achieving improved uptake in treatment for stigmatized disorders exist. For example, we have found improved understanding of depression, decreased stigma, and increased treatment engagement after an education intervention (52). Similarly, interventions focused on education campaigns (58), leaflets with photographs (59), and motivational interviewing (60) show promise. Just the receipt of mental health services, including psychotherapy, has led to dispelling stigmatizing attitudes (61).

Limitations of this study are present in the sample utilized, analyses conducted, and structure of the measures. First, the sample was recruited via a purposive, non-probability sampling method, producing a relatively homogenous group of treatment-seeking, primarily Spanish-speaking females of Hispanic origin, likely Mexico. As such, the results may not be generalizable to other Hispanic populations including those that are racially mixed (such as Afro-Hispanics), male, English speaking, or not seeking mental health treatment. This homogenous nature of the sample as well as a lack of data on other important sociocultural factors not captured in the medical record (such as nativity, immigration status, acculturation, and Hispanic subgroup) also limited our ability to test sources of invariance or the effects of covariates within the models. Also, due to the secondary nature of the current study, additional psychometric analyses, such as test-retest reliability and criterion-related validity, could not be performed. In terms of the measures themselves, the 3-item nature of the SCMHC provided for a statistical limitation in being able to adequately assess model fit. Finally, the response option of “don’t know” led to a particularly high rate of missing data for items in the LSAS, which may indicate problems with measure comprehension or limited willingness to answer questions specific to antidepressant medication among participants. Despite these limitations, this study addressed an important gap in the literature regarding the validation of stigma measures for Hispanics using a large sample of primary care patients who have been diagnosed with depression.

Assessing for barriers to depression treatment, including stigma, are critical to understanding engagement of racial and ethnic minority populations in care and eliminating disparities. Although our ability to assess model fit for the SCMHC was limited, this measure shows promise as a useful tool in both clinical and research settings given its ease of use and significant relationship to mental health treatment utilization outcomes (8). Future studies should continue to evaluate validated stigma measures to understand the impact of stigma on mental health care engagement and ways in which interventions can be developed and implemented to reduce stigma and increase treatment uptake among Hispanics. Furthermore, future research should include more heterogeneous samples within Hispanic populations and the testing of covariates to provide more information on the usefulness of measures among different subgroups. The addition of a stigma checklist and simple strategies to counteract stigma to the clinical dialogue between patient and provider could improve rapport and help overcome barriers to treatment. The limitations of the primary care context such as appointment time, provider training, patients’ competing physical disorders and lack of reimbursement remain a challenge.