Research Article

Culturally Adapted Psychotherapeutic Interventions for Latino Depression and Anxiety: A Meta-Analysis

Abbie Nelson1 , Esther Ayers1, Fei Sun1, and Anao Zhang2

Abstract Objectives: Culturally adapted psychotherapeutic interventions have been developed to treat Latino depression and anxiety. Evidence is lacking regarding the overall effectiveness and generalizability of these adapted interventions. This study conducted a systematic review and meta-analysis of psychotherapeutic interventions for Latino depression or anxiety. Method: A search of nine electronic databases and manual review of reference lists were conducted. Thirteen studies of randomized controlled trials were eligible and included for meta-analysis using robust variance estimation in meta-regression. Results: An overall small treatment effect that is statistically significant was identified for Latino depression or anxiety of d¼ 0.334, 95% confidence interval [0.049, 0.619], p < .05. Discussion and Implications: More research is needed to determine the effectiveness of culturally adapted psychotherapeutic interventions for Latino depression or anxiety.

Keywords Latino, psychotherapeutic interventions, Latino depression or anxiety, systematic review, meta-analysis

Anxiety and depression, affecting all demographics across the

life span, are the two most common mental health disorders for

the population in the United States, with anxiety disorders

estimated at 28.8% (Kessler et al., 2005) and depression esti-

mated at 17% (Centers for Disease Control and Prevention

[CDC], 2018). An estimated 7.1% of children and teenagers

(CDC, 2019) and 21.7% of adults (Kessler et al., 2005) in the

United States have an anxiety disorder in their lifetime. One in

six Americans, approximately 17%, will experience depression

in their life span (CDC, 2019). The estimates of overall pre-

valence rates of depression among Latinos were higher than the

national average of 17% at a rate of 27% determined by a cross-

sectional analysis of 15,864 men and women aged 18–74 years

in the population-based Hispanic Community Health Study/

Study of Latinos (Wassertheil-Smoller et al., 2014). In this

study, a shortened Center for Epidemiological Studies Depres-

sion Scale was used to assess depression (Wassertheil-Smoller

et al., 2014). Alegria et al. (2008), found a rate of anxiety

disorders for U.S.-born Latinos 18.9% compared to foreign-

born Latino immigrants at a lower rate of 15.2% using data

from the National Latino and Asian American Study and the

National Comorbidity Survey Replication (Kessler & Merikan-

gas, 2004). Although numbers vary depending on the study and

method, it is still clear Latinos are experiencing anxiety and

depression. We used Latino to refer to Hispanic and Latino

population studied in this research. The term “Hispanic”

emerged in the middle to late 1970s and was used by the Cen-

sus Bureau in the 1980s to describe people with varying

backgrounds but share a common language and cultural heri-

tage (Del Olmo, 2001 as cited by Delgado, 2007). The term

Latino emerged in the early 1990s and it is not uncommon to

see the terms Hispanic and Latino used interchangeably (Del-

gado, 2007).

The Latino population in the United States was 56.5 million

in 2015 accounting for 17.6% of the total population, and it is

expected to grow to 29% of the population by 2050 (Pew

Research, 2015).The high prevalence of both depression and

anxiety among Latinos and the limited number of culturally

adaptive interventions and research of efficacy create the need

for research to identify if culturally adaptive interventions are

effective in addressing depression and/or anxiety among Lati-

nos. Without effective treatment for Hispanics with anxiety and

depression, the negative societal effects will compound as the

percentage of the population grows likely leading to more eco-

nomic, health, and familial costs to individuals and society.

This study reports findings of a systematic review and meta-

analysis of culturally adapted psychotherapeutic interventions

for treating Latino depression or anxiety. The purpose of this

1 School of Social Work, Michigan State University, East Lansing, MI, USA 2 School of Social Work, University of Michigan, Ann Arbor, MI, USA

Corresponding Author:

Abbie Nelson, School of Social Work, Michigan State University, 655 Audi-

torium Rd #254, East Lansing, MI 48824, USA.

Email: [email protected]

Research on Social Work Practice 2020, Vol. 30(4) 368-381 ª The Author(s) 2020 Article reuse guidelines: sagepub.com/journals-permissions DOI: 10.1177/1049731519899991 journals.sagepub.com/home/rsw

study is to evaluate whether culturally adapted psychothera-

peutic interventions are effective for treating Latino depression

or anxiety. Studies included had interventions for treating

depression or anxiety because depression and anxiety are

highly comorbid, and both outcomes are often reported simul-

taneously in clinical trials.

Compared with non-Latino Whites, Latinos with similar

mental health needs are using mental health services less due

to cultural, structural, and economic factors (Cabassa et al.,

2006). Other barriers to receiving the care they need are lack

of Latino health-care professionals, low socioeconomic levels,

higher rates of uninsured, and societal and individual preju-

dices and discrimination (Ruiz, 2002). If mental illness is left

untreated, it negatively impacts daily functioning causing poor

concentration (Aisenberg et al., 2012), lack of energy (Alegria

et al., 2008), can result in a poorer response to future treat-

ments, and a higher risk the mental illness will become chronic

(Ghio et al., 2013).

Despite the critical need for services, research continues to

identify mental health service gaps among Latinos and suggests

that the use of nonculturally sensitive interventions may not be

effective for Latino populations. The use of culturally adapted

interventions may be a solution for encouraging adherence and

retention to mental health services for Latinos.

Cultural match theory states individuals benefit from inter-

ventions that align more closely with their cultural character-

istics (La Roche et al., 2011). La Roche et al. tested this theory

with the Latino population and found patients were more likely

to adhere to a culturally competent relaxation intervention that

aligned with their cultural values of allocentrism, a term used to

describe a personality trait in which a person focuses their

attention on others’ actions more than themselves. Specifically,

for depression, a culturally adapted behavioral activation tech-

nique showed preliminary success in Latino engagement and

continued investment in services (Kanter et al., 2010). In recent

years as the importance of culturally competent treatment mod-

alities emerged, conceptual models and frameworks for imple-

menting cultural adaptations have been created and

implemented with the realization of the complexity of this

process. Bernal et al. (2009) highlighted the different views

of treatments, one view stating universal treatments should

be able to apply to all people, while others called for a com-

pletely separate approach to interventions depending on the

culture (Comas-Diaz, 2006). As a middle ground between the

above approaches, culturally adapted evidence-based treat-

ments show positive results with diverse populations such as

Latinos and Asian Americans. However, efficacy is inconclu-

sive due to the various ways of measuring cultural adaptations.

For behavioral health interventions, there has been enough

literature published on steps to take to culturally adapt inter-

ventions that five common stages have emerged that behavioral

health researchers and practitioners can use when determining

when and how to implement adaptations (Barrera et al., 2013).

All of the above frameworks give great guidance on how to

identify the need and implement cultural adaptations. A com-

prehensive review article (Jani et al., 2009) looking at the effect

of cultural adaptations of interventions used in the areas of

health, mental health, and substance abuse with Latinos found

that most of the 23 studies included showed positive outcomes

after implementing a culturally adapted intervention. The

groundwork has been laid, but more research is needed to

synthesize the evidence and examine the effect size of cultu-

rally adapted interventions on client outcomes.

Limited systematic reviews attempted to conclude the effec-

tiveness of the statistical significance compared to treatment.

One systematic review study (Pineros-Leano et al., 2017)

examined 11 randomized controlled trial (RCT) and non-

RCT studies to determine the effectiveness of cognitive beha-

vioral therapy (CBT) with immigrant Latinos and described the

various cultural adaptations. Pineros-Leano et al. concluded

that the CBT interventions resulted in decreased depression

symptoms. They described the cultural adaptations used in the

various studies such as environmental and cognitive adapta-

tions but were not able to measure how effective these were.

Another systematic review of 36 RCT and non-RCT studies

focused on depression treatment among Latino adults (Collado

et al., 2016). Thirty-five of the studies incorporated cultural

modifications to the interventions, with the most common

adaptation being the provision of therapy in Spanish. CBT

delivered in client’s homes or via teletherapy showed signifi-

cant reductions in depression compared to control participants.

Individual outcomes had the best results when using interper-

sonal treatment for 16 sessions. The study called for a more

thorough evaluation of treatment moderators relevant to Lati-

nos, including language preference, acculturation, and subse-

quent case management given the heterogeneity of the

population. With the results of the above studies, it remains

unclear how effective cultural adaption is for this population

and what components are responsible for this effect if there is a

significant one.

Even fewer meta-analyses examined the effect of culturally

adapted interventions, without a focus solely on Latinos. A

meta-analysis by Escobar and Gorey (2018) sought to deter-

mine whether culturally adapted cognitive behavioral interven-

tions (CBIs) have a different effectiveness than those without

adaptation. They determined the effect to be significant at post-

intervention period, d ¼ 0.41, 95% confidence interval (CI)

[0.30, 0.52], and at 6- to 12-month follow-up with d ¼ 0.44,

95% CI [0.30, 0.58] when compared to interventions that were

not adapted or were only adapted on “surface structure,” which

involved matching the intervention to observable characteris-

tics of the population (e.g., language) but did not include

deeper cultural aspects. The authors suggested the continued

need for research due to small sample sizes and the RCTs being

more randomized pilot trials that lacked blinding. Another

meta-analysis (Van Loon et al., 2013) looked at the effective-

ness of culturally adapted depression and anxiety treatments for

ethnic minorities in Western countries and found a significant

pooled effect size of 1.06, 95% CI [0.51, 1.62], p < .001. They

contributed the effectiveness to the adaptations on cultural val-

ues and beliefs related to the healing process. Of the nine

studies used for Van Loon et al.’s review, only three focused

Nelson et al. 369

specifically on the Latino population showing the knowledge

gap of the effectiveness specifically for Latinos.

As evidenced by the above literature, no meta-analyses have

focused on culturally adapted interventions using RCTs to treat

anxiety or depression among Latinos. While culturally adapted

intervention research for Latinos tends to grow, research is

needed to determine whether and how culturally adapted inter-

ventions are working and which conditions among Latinos

yield the most effectiveness. In addition, conclusions taken

from previous studies that lacked the rigor of RCTs may be

producing biased results. Therefore, this study will add to the

growing knowledge base in an important way.

Method

Following the Cochrane Collaboration Guidelines (Higgins &

Green, 2011), this study used various search strategies to obtain

relevant literature published between 1900 and February 2019.

The strategies included searching nine electronic databases,

grey literature, and reference lists in related systematic reviews.

The nine electronic databases were CINAHL Plus with Full

Text, Family & Society Studies Worldwide, Gender Studies

Database, Social Sciences Full Text (H.W. Wilson), Education

Resources Information Center, Academic Search Complete,

Health Source: Nursing/Academic Edition, Psychology and

Behavioral Sciences Collection under EBSCOhost, and Sco-

pus. These databases were included as they covered a wide

range of journals related to mental health and the topic of

interest to this review. The initial eight databases were searched

using the search terms to identify culturally adapted

(“culturally adapted” or “culturally competent” or “culturally

sensitive”) AND Latino population (latin*) AND therapy

(therap*) AND depression (depress*). Choosing only peer-

reviewed articles with these terms yielded only 35 results. The

search was then expanded by using the search terms to identify

culturally adapted (cultur*) AND Latino population (latin*)

AND therapy (therap*) AND (depress*) OR anxiety (anx*),

which yielded 170 results. The same broader terms were then

used in Scopus, which added 126 results. After input from the

reviewer, we did post hoc analyses using the term Hispanic and

resulted in an additional four studies.

Inclusion and Exclusion Criteria

To be eligible for inclusion, a study needed to be (1) an RCT

that took place in the United States and (2) examining the

effects of psychotherapeutic intervention on depression and/

or anxiety of Latinos. Study participants could be of any age

as long as they were Latino. Psychotherapeutic interventions

are broadly defined in this review to include therapeutic tech-

niques or strategies used by therapists and psychosocial inter-

ventions implemented by community health practitioners with

the intent to improve mental health symptoms. Community

health practitioners are persons with knowledge about the com-

munity they work in and are providing support, services, or

psychosocial interventions with the desire to create positive

change in the community. Community health workers such

as promotoras referring to community health workers in

Spanish-speaking communities are also included in this cate-

gory (Office of Minority Health & Health Equity, 2019). Lati-

nos were operationalized as a person of any age belonging or

relating to a culture from Latin America or other countries that

speak Spanish/and or English and define themselves as Latinx.

Culturally adapted was defined as any item of the intervention

modified to be sensitive to the culture of individuals who are

receiving the treatment that takes into account their values,

language, rituals, social networks, background, and “lived

experience of the participants” (Marsiglia & Booth, 2015).

Culturally adapted interventions for Latinos may include but

are not limited to including the following features: program

delivered in Spanish, use of Latino community health workers

“promotora,” and addressing mental health beliefs and norms

specific to Latino community. A study was included if it used

measure(s) of depression and/or anxiety as either primary or

secondary outcomes. For example, a study that examined the

effect of a culturally sensitive cognitive behavioral group inter-

vention for Latino Alzheimer’s caregivers and measured neu-

ropsychiatric symptoms as well as depression would be

included. Studies were only accepted if published in English.

A study would be excluded if (1) Latinos were not a part of the

intervention, (2) did not have a randomly assigned control/

comparison group, (3) did not contain measure(s) of depression

and/or anxiety, (4) did not report necessary statistical informa-

tion for effect size calculations, (5) did not include a cultural

adaptation to the intervention, and (6) was conducted in another

country.

Screening and Data Extraction

Two doctoral-level students and one postdoctoral fellow parti-

cipated in eligibility screening, and the two doctoral students

completed the data extraction procedures. Using Covidence

online software (https://www.covidence.org/home), title and

abstract of each study was screened by each doctoral student

independently and blinded to each other’s decisions. The third

independent screener (postdoc fellow) resolved the conflicts.

During full-text screening, the two doctoral students indepen-

dently reviewed the full texts and discussed any differences.

The independent reviewer was consulted on the articles that the

students could not find a consensus. A manual review was

conducted of the reference lists of systematic reviews related

to interventions for Latinos for depression or anxiety to identify

any missed studies from the search for inclusion. A coding

sheet for data extraction was developed and then used to guide

coding of all included studies.

Coding domains (available upon reasonable request from

the first author) consisted of participant and provider charac-

teristics, intervention characteristics, research design, and

effect size data. Participant characteristics included age, gen-

der, marital status, socioeconomic status, and Latino subgroup.

Provider characteristics included profession, educational back-

ground, clinical experience, and whether they received

370 Research on Social Work Practice 30(4)

supervision and training. Intervention characteristics included

an intervention’s type (e.g., therapeutic vs. supportive), dosage

(minutes per session, number of sessions, and duration in

weeks), format (e.g., individual, group), delivery methods

(e.g., in-person, technology-assisted), and delivery setting

(e.g., home, community-based mental health service).

Research design included the nature of the control group, diag-

nostic tools, and outcome measure(s). Since all studies

included were RCTs, the nature of the comparison group was

coded as treatment as usual or waitlist control.

Interscreener and Interrater Agreement

Satisfactory interscreener reliability was observed with 79% for title/abstract and 85% for full-text screening. Interrater

reliability was calculated using a percent agreement model

by dividing the number of agreements over all possible extrac-

table data points. The two researchers reached a satisfactory

interrater agreement of 87% for data extraction. All disagree-

ments were resolved by discussion and consensus.

Data Analysis

Data extraction was conducted in Microsoft Excel and ana-

lyzed using R software (version 1.1.463) in four stages: (1)

conducting descriptive statistics of participants, providers,

intervention characteristics, and research design; (2) calculat-

ing small sample–corrected effect size estimates; (3) synthesiz-

ing effect size estimates across studies; and (4) moderator

analysis using meta-regression.

Effect Size Calculation

All of the depression and anxiety outcomes reported in the

studies for this analysis were continuous outcomes. There-

fore, their effect size estimates were calculated using Hedges’

g effect size (Cooper et al., 2009). Hedges’ g represents stan-

dardized mean differences when different measures were used

across studies. The g value was further adjusted using

Hedges’ small sample size correction (Hedges, 1981) to

obtain an unbiased estimate. This estimate is noted as “d” in

this review.

Effect Size Synthesis and Moderator Analysis

An overall treatment effect size estimate and moderator anal-

yses were conducted using robust variance estimation (RVE) in

meta-regression, a method that has been supported by both

methodological and previous empirical studies (Hedges et al.,

2010; Tanner-Smith & Tipton, 2014; Zhang et al., 2019). Sev-

eral studies included in this review reported multiple effect

sizes, which introduced statistical dependence into the resulting

effect sizes. Comparing with other statistical procedures that

handle within-study dependence, such as generalized least

squares estimation (Olkin & Gleser, 2009) or multilevel

meta-analysis modeling (Van den Noortgate et al., 2013), RVE

fits better because RVE makes no assumptions about effect size

sampling distributions and can estimate the covariance struc-

ture of the dependent effect sizes without statistical information

needed in other methods. Methodological studies recom-

mended an ideal sample size of five effect sizes per study and

40 studies for RVE to generate reasonably accurate results

(Hedges et al., 2010; Tipton, 2015). Because existing studies

meeting the inclusion criteria of this study is less than 40, we

used small sample size adjustment when running the meta-

regression with RVE (Tipton, 2015).

Having identified variability among effect size estimates,

we conducted moderator analyses to examine factors that influ-

ence effect size. Moderator analysis is able to indicate the

statistical differences between different subgroups. Since it

does not determine if the effect size of each subgroup is statis-

tically significant, an analysis of treatment effect among sub-

groups is conducted alongside moderator analysis to improve

the results of clinical significance.

Publication Bias and Risk of Bias

Publication bias refers to the nonrepresentativeness of articles

published among all the research done in an area. This happens

when studies with nonsignificant results are less likely to be

published, which affects the bias of effect sizes. This review

used Vevea and Woods’ (2005) weight function model to sta-

tistically assess the possibility of publication bias, and a funnel

plot of the effect size estimates graphed against their standard

errors was used to visually depict the bias. Risk of bias was

assessed in light of the evaluation criteria specified in the

Cochrane Handbook for Systematic Reviews of Intervention

(Higgins & Green, 2011).

Results

Search Results

The Preferred Reporting Items for Systematic Reviews and

Meta-Analyses (PRISMA) diagram in Figure 1 shows the

results of a detailed literature search and the results from the

selection process. A total of 331 references were uploaded to

Covidence for screening, 98 duplicates were removed leaving

233 for title and abstract screening. Of the 233 articles, on