The methods and outcomes of cultural adaptations of psychological treatments for depressive disorders: a systematic review

Background Cultural adaptations of evidence-based psychological treatments (PTs) are important to enhance their universal applicability. The aim of this study was to review systematically the literature on adaptations of PTs for depressive disorders for ethnic minorities in Western countries and for any population in non-Western countries to describe the process, extent and nature of the adaptations and the effectiveness of the adapted treatments. Method Controlled trials were identified using database searches, key informants, previous reviews and reference lists. Data on the process and details of the adaptations were analyzed using qualitative methods and meta-analysis was used to assess treatment effectiveness. Results Twenty studies were included in this review, of which 16 were included in the meta-analysis. The process of adaptation was reported in two-thirds of the studies. Most adaptations were found in the dimensions of language, context and therapist delivering the treatment. The meta-analysis revealed a statistically significant benefit in favor of the adapted treatment [standardized mean difference (SMD) −0.72, 95% confidence interval (CI) −0.94 to −0.49]. Conclusions Cultural adaptations of PTs follow a systematic procedure and lead primarily to adaptations in the implementation of the treatments rather than their content. Such PTs are effective in the treatment of depressive disorders in populations other than those for whom they were originally developed.


Introduction
Although there is extensive evidence of the effectiveness of psychological treatments (PTs) for depressive disorders (Cuijpers et al. 2008;Hollon & Ponniah, 2010), it has been argued that PTs are developed in particular cultural contexts and that this may limit their universal applicability. Adapting evidence-based PTs to incorporate elements that are contextually relevant and meaningful in the culture in which they are being delivered is recognized as an important step to increasing acceptability of the treatment, patient satisfaction and, ultimately, their effectiveness (Bernal & Scharrón-del-Río, 2001;Sue, 2003;Castro et al. 2010). Cultural adaptations of PTs can be viewed as a middle ground between the two extreme positions of considering an original evidence-based intervention as applicable to all cultural groups without the need for adaptation and a culture-specific approach with emphasis on unique culturally grounded content and process. This middle ground ensures that a cultural adaptation attempts to maintain fidelity to the core elements of the PT while adding certain cultural elements to enhance its acceptability and effectiveness (Falicov, 2009;Barrera et al. 2013). Understanding the process by which researchers have made these adaptations and the specific nature of the cultural adaptations may serve to inform others interested in tailoring PTs to specific populations of diverse cultures.
Although there have been previous systematic reviews of adaptations of mental health interventions (Griner & Smith, 2006;Huey & Polo, 2008;Benish et al. 2011;Smith et al. 2011), these have been heterogeneous in their scope (for example, addressing multiple mental health problems) and have left undescribed the process or nature of the adaptations. Notably, although there is a growing body of evidence supporting the effectiveness of adapted PTs for depressive disorder in non-Western countries, many of which have been adapted to the local context Rahman et al. 2008;Patel et al. 2011), these have not been included in existing reviews. The aims of this review were to synthesize the literature on cultural adaptations of PT for depressive disorder for both ethnic minorities in Western countries and culturally diverse populations in non-Western countries in three respects: (1) to describe the procedures used to adapt the PT; (2) to describe the extent and nature of the adaptations to the PT; and (3) to assess the effectiveness of the adapted PT.

Method
We used standard methods for systematic reviews and meta-analyses in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement (Moher et al. 2009).

Identification of studies
Studies were identified by a systematic literature search using the following strategies: (1) A database search of Ovid Medline, EMBASE and PsycINFO until December 2011 was conducted to identify controlled trials conducted in Western countries with ethnic minority populations. The search terms used are described in Appendix 1 in the online Supplementary material. For studies from non-Western countries, the terms for ethnic minority populations were replaced with individual country names (i.e. World Bank-defined lowand middle-income countries as these would ensure inclusion of mainly non-Western countries). No start date was specified.
(2) Cross-referencing of eligible articles to identify additional studies that met our inclusion criteria. (3) Key informants (i.e. known PT experts, including authors of the eligible studies) were contacted to identify other studies that could be included in our review. (4) Bibliographies of key reviews of PTs (Patel et al. 2009) were hand searched to identify studies that may have been missed through the database search.

Inclusion criteria
(1) Randomized controlled trials (RCTs) or non-RCTs that described the evaluation of PTs for depressive disorder in an ethnic minority population in a Western country or any evaluation of a PT in a non-Western country. Non-RCTs were included as they too would provide information on all the research questions including effectiveness.
(2) No restriction on language, sample size, type of comparison group or outcome measure. (3) Studies conducted in adults (aged 519 years) with depressive disorder.

Exclusion criteria
(1) Studies that adapted the PT only to facilitate access (e.g. home delivery of PT) rather than to address issues of broader cultural relevance.
(2) Studies that used a new PT that was developed specifically for the ethnic group and was not an adaptation of an existing PT.

Data collation and extraction
The titles and abstracts of each citation identified from our search were inspected independently by two reviewers (N.C. and A.N.) with reference to the inclusion and exclusion criteria and to identify duplicates. The potentially relevant full-text papers were accessed and reviewed independently by the two reviewers. Any disagreements were resolved by consensus and, when this could not be reached, a third reviewer (V.P.) adjudicated. To address the first and second questions of the review, papers that referenced previous publications describing the adaptation process and details of adaptations were also retrieved. Corresponding authors of all papers were contacted to retrieve any additional information regarding these two questions. The questionnaire asked authors to verify the accuracy of information extracted and to add any information regarding process and adaptation that were missing in the papers. Authors were also asked to describe aspects of the PTs that did not require adaptation. Data were summarized in a table based on the theoretical frameworks described below. The quality of included studies was assessed on the following criteria: method of randomization, allocation concealment, blinding of outcome assessment and attrition bias.

Data analysis
Thematic analysis was used to evaluate the process and nature of the adaptations. Analysis was deductive at first, consisting of predetermined themes applied to data. These themes were based on two frameworks: (1) The Medical Research Council (MRC) framework for the development and evaluation of complex interventions was used to draw out common elements in the process of cultural adaptation (Craig et al. 2008). This framework recommends a phased development process consisting of: modeling/theoretical development, formative work, piloting and evaluation. We categorized the process of adaptation described in the studies into these steps.
(2) Several frameworks were considered for analysis of the nature of the cultural adaptations (Lau, 2006;Hwang, 2009;Castro et al. 2010) and the framework of Bernal & Saez-Sanriago (2006) was selected. This framework includes eight dimensions that can be the targets of cultural adaptations: (1) language of the intervention, (2) therapist matching, (3) cultural symbols and sayings (metaphors), (4) cultural knowledge or content, (5) treatment conceptualization, (6) treatment goals, (7) treatment methods, and (8) consideration of treatment context. Subsequent analysis was inductive and focused on the generation of new categories for the details of the cultural adaptations that emerged from the data. Meta-analysis was performed using Review Manager version 5.1. The included studies assessed depression (primary outcome) using different psychometric scales; therefore, standardized mean differences (SMDs) were used as appropriate.
The I 2 test was used to measure statistical heterogeneity across studies. A random-effects model was used for the meta-analyses because substantial heterogeneity was observed (I 2 = 90%) (Higgins et al. 2003). The uncertainty around heterogeneity was explored with subgroup analyses. A funnel plot was charted to    (Egger et al. 1997); the plot was asymmetrical suggesting publication bias.

Process of cultural adaptation
The application of the MRC framework to the selected studies is shown in Table 2. Six studies (32%) systematically followed all four stages of the MRC framework in their adaptation process Rahman et al. 2008;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Patel et al. 2011). The remaining studies either did not describe the adaptation process in sufficient detail or omitted some of the stages in treatment development.
Most authors described the need for cultural relevance to increase acceptability of the PT as the primary rationale for carrying out the adaptation. Other reasons were to address practical barriers to care, such as limited literacy and lack of trained mental  Eleven studies (58%) reported the use of the modeling stage (Kohn et al. 2002;Bolton et al. 2003;Miranda et al. 2003b;Patel et al. 2003Patel et al. , 2011Rahman et al. 2008;  Grote et al. 2009;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Naeem et al. 2011). This involved the selection of a theory-based class of PT, such as cognitive behavior therapy (CBT), mainly through empirical evidence gathered from literature reviews and also through consultations with local stakeholders, for example mental health specialists, primary care doctors and community workers. The formative research phase using mixed methods to guide the process of development of the preliminary PT manual was reported in eight (42%) studies (Araya et al. 2003;Bolton et al. 2003;Rahman et al. 2008;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Naeem et al. 2011;Patel et al. 2011). This consisted mainly of collection of qualitative data through focus group discussions and in-depth interviews with the target group (i.e. patients with depressive disorder) and specialist and non-specialist health providers with experience of working with the specific cultural groups or belonging to the same culture (hence having knowledge of the customs/beliefs of that culture). These data were used to refine the treatment's acceptability and feasibility.
Piloting, leading to further refinement of the treatment, was reported in 10 (53%) studies (Kohn et al. 2002;Bolton et al. 2003;Patel et al. 2003Patel et al. , 2011Rahman et al. 2008;Grote et al. 2009;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Naeem et al. 2011). This involved training the health-care providers to deliver the preliminary version of the treatment and the collection of quantitative and qualitative data either in open trials or in controlled trials with small numbers of patients to improve the acceptability and feasibility of the treatment. Pilot studies also provided an initial assessment of the ability of the treatment to change depression outcomes.
The evaluation phase comprised a controlled trial of the PT comparing the adapted treatment to a control condition in RCTs or non-RCTs, the results of which are described in the following sections.

Cultural adaptations of PT
The types of PT adapted consisted of CBT (n = 10), interpersonal therapy (IPT; n = 4), psychoeducation (n = 3), problem-solving therapy (n = 2) and dynamically oriented therapy (n = 1). Details of the adaptations made to the PT, according to type of PT, based on the framework of Bernal & Saez-Sanriago (2006) are presented in Appendix 3 online. Ten authors completed the questionnaire sent to them. All the studies reported adherence to the basic framework and core principles (e.g. the four IPT problem domains) of the original PT so as to preserve fidelity to the treatment. In addition, the authors identified several aspects of the PT that did not require adaptation. These included the phases in which the treatment was delivered, the use of problem-solving techniques and the empathic nature and other 'non-specific' aspects of the therapeutic relationship.
Therapist adaptations, found in 14 (70%) studies (Comas-Diaz, 1981;Araya et al. 2003;Bolton et al. 2003;Miranda et al. 2003b;Patel et al. 2003Patel et al. , 2011Rahman et al. 2008;Grote et al. 2009;Afuwape et al. 2010;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Dwight-Johnson et al. 2011;Naeem et al. 2011), focused on therapist-patient matching to ensure the acceptability and credibility of the therapist by emphasizing shared experiences and awareness of local customs. Some studies reported specific training of therapists in cultural competence to enhance patient engagement. Additionally, cultural factors were considered in the therapist-patient relationship; for example, the therapist adopting a less directive style in some studies and the necessity of setting therapeutic boundaries in others.
The use of metaphors to increase cultural relevance was reported in six (30%) studies (Crespo, 2006;Rahman et al. 2008;Grote et al. 2009;Dwight-Johnson et al. 2011;Naeem et al. 2011;Patel et al. 2011). These took the form of using material that was culturally relevant, for example a health calendar to monitor homework, the use of stories and local examples with characters resembling the patient's situation and background, and the use of idioms and symbols such as beads for counting thoughts and a mood ladder for rating mood.
Cultural considerations were integrated into the content of the PT in eight (40%) studies (Kohn et al. 2002;Bolton et al. 2003;Patel et al. 2003;Wong, 2008b;Grote et al. 2009;Ell et al. 2010;Naeem et al. 2011;Patel et al. 2011). These took the form of addressing stressful circumstances such as interpersonal difficulties and focusing on what was in the patient's control. Furthermore, local remedies and practices were integrated into the treatment; for example, massage and religious therapy and additional modules (such as on spirituality) were added to the PT manual, when needed, to contextualize the treatment and address issues relevant to the cultural group.
Adaptations in the dimension of concepts, described in six (30%) studies (Comas-Diaz, 1981;Bolton et al. 2003;Wong, 2008b;Grote et al. 2009;Beeber et al. 2010;Naeem et al. 2011), involved the communication of the presenting problem and its constructs in a culturally appropriate manner so that they are understandable by the patient and reduced stigma. In particular, this involved addressing the somatic conceptualization of depression by avoidance of psychiatric labels and presenting the problem as a medical illness rather than 'madness'.
Goals were the least commonly modified dimension, described in four (20%) studies Rahman et al. 2008;Grote et al. 2009;Naeem et al. 2011). Adaptations involved development of clientderived treatment goals that were personally and culturally relevant, such as focusing on the health of the family unit rather than the individual. Goals were also extended beyond depression treatment, for example by enhancing roles of group members into community advocates.
Finally, 13 (65%) studies reported adaptations to ensure that the PT fit into the patient's broader social context (Comas-Diaz, 1981;Dai et al. 1999;Kohn et al. 2002;Bolton et al. 2003;Miranda et al. 2003b;Grote et al. 2009;Hamdan-Mansour et al. 2009;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Dwight-Johnson et al. 2011;Naeem et al. 2011;Patel et al. 2011). These took the form of adaptations to reduce practical barriers and improve access, enhance feasibility and acceptability; for example, flexibility in scheduling sessions, delivering the treatment in a convenient setting or over the telephone, and inclusion of family members if requested by the patient.

Effectiveness of adapted PTs
The primary outcome, depressive symptoms, was measured using nine different depression scales, most commonly the Beck Depression Inventory (BDI) (n = 5); other scales are shown in Table 1. The duration of the follow-up ranged from 2 weeks to 18 months post-treatment. All but two studies (Patel et al. 2003;Gater et al. 2010) reported the effectiveness of a culturally adapted intervention in depression (see Table 1).
Sixteen of the 20 studies with 4162 participants were included in the meta-analysis Miranda et al. 2003b;Patel et al. 2003Patel et al. , 2011Rojas et al. 2007;Rahman et al. 2008;Wong, 2008b;Grote et al. 2009;Hamdan-Mansour et al. 2009;Afuwape et al. 2010;Beeber et al. 2010;Ell et al. 2010;Gater et al. 2010;Dwight-Johnson et al. 2011;Naeem et al. 2011). Four studies were not included in the meta-analysis: three (Comas-Diaz, 1981;Dai et al. 1999;Kohn et al. 2002) did not provide sufficient information on outcome measures (i.e. they did not report the mean and standard deviation in the two groups, which is necessary to calculate the SMDs) and one (Crespo, 2006) was a poor quality study that administered active (i.e. non-adapted) PT to the comparison group. The pooled weighted SMD showed a statistically significant benefit in favor of the adapted PT over the various control conditions [SMD = −0.72, 95% confidence interval (CI) −0.94 to −0.49], with significant heterogeneity (χ 2 = 146, df = 15, p < 0.001, I 2 = 90%) (Fig. 2). This indicates that culturally adapted PTs are efficacious.
Although some of the subgroup analyses revealed differences in effect sizes, none were statistically significant (Table 3). Moreover, significant heterogeneity was observed for each subgroup analysis (I 2 > 50%).
Of note, two studies compared the effectiveness of the culturally adapted intervention to the non-adapted intervention directly (Kohn et al. 2002;Crespo, 2006). Although both studies reported greater effectiveness of the adapted to the non-adapted PT, we were unable to obtain an integrated effect size because the results did not provide sufficient statistical information to allow this calculation.

Discussion
In this study we sought to systematically review the literature on adaptations of PTs for depressive disorders for use with ethnic minorities in Western countries and any adaptations of PTs for depressive disorder in non-Western countries. We have used the terms Western/non-Western to reflect the difference between populations in which the psychological interventions were originally developed and tested (always a white European or American population) versus populations with different concepts of health and illness.
We identified 20 controlled studies of PTs in these populations. The process of adaptation was reported explicitly in approximately two-thirds of the studies. The common elements within this framework were the selection of a theory-driven class of PT, consultation with a variety of stakeholders in the adaptation process, the use of mixed (qualitative and quantitative) research methods to assess acceptability and feasibility   Fig. 2. Effect of psychological interventions compared to usual care/no treatment control group. Outcome: depression (higher score indicates greater severity). CI, confidence interval; df, degrees of freedom; S.D., standard deviation. of the PT and pilot studies to evaluate barriers to the delivery of the PT, before the treatment was evaluated in a controlled study (which was an inclusion criterion for the review). Our findings confirm that the procedure used by most investigators is consistent with the methodological framework for the development of complex interventions advocated by the MRC (Craig et al. 2008). Several themes emerged when examining the nature of adaptations to the PT, which were organized using the framework of Bernal & Saez-Sanriago (2006). The use of this framework offers a systematic basis for identifying the various dimensions that need attention in cultural adaptations of PTs. The majority of adaptations were made in the dimensions of language, context and therapist delivering the treatment. Replacing technical terms with colloquial expressions, ensuring therapist-patient matching and cultural competence of therapists were some of the important adaptations reported. This is similar to a previous meta-analysis that reported matching patient to therapist of the same ethnic group in 61% of studies and the same native language (if other than English) in 74% of studies (Griner & Smith, 2006). Other adaptations of salience are the incorporation of local practices into treatment, extending the goal of treatment beyond the patient to include the family, attention to the somatic/physical illness model and simplification of treatment including the use of non-written material. It is important to note that many aspects of the PTs were found to be 'universally' applicable, that is they did not require adaptation and the framework and theory of the treatment including treatment phases remained unchanged, respecting the theoretical core of the original PT. Thus, adaptations predominantly reflected efforts to enhance the acceptability of the PTs as opposed to adaptations of core content, thus maintaining fidelity to the original PT.
Despite significant heterogeneity, the meta-analysis of 16 studies confirmed the large effect size of adapted PTs (SMD −0.72). The heterogeneity in our review is not unexpected given the diversity of contexts and PTs in the studies included (Higgins et al. 2003). The pooled effect sizes of adapted treatments were similar or greater than those reported in systematic reviews of the non-adapted treatment in 'Western populations'; for example the effect size of adapted CBT (n = 6) (SMD −0.88, 95% CI −1.29 to −0.48) was similar to the effect sizes of non-adapted CBT reported in two reviews [SMD −0.33, 95% CI −0.60 to −0.06 (Peng et al. 2009) and SMD −1.34, 95% CI −1.89 to −0.79, p < 0.00 001 (Cape et al. 2010)] whereas the effect size of adapted IPT (pooled SMD −0.97, 95% CI −1.62 to −0.31) was greater than the non-adapted IPT (SMD −0.11, 95% CI −0.47 to 0.24) (Cape et al. 2010). This suggests that adapting PTs for culturally diverse populations can be achieved without compromising treatment effectiveness, and indeed may potentially enhance it.
Of particular interest is the comparative effectiveness of an adapted treatment with a non-adapted treatment in the same population. One previous systematic review (Benish et al. 2011) compared adapted PTs with non-adapted PTs (21 studies) and reported a modest effect size (d = 0.32) in favor of culturally adapted PTs. That review, however, included studies that were not eligible for our current review because the PTs were for a wide range of psychiatric disorders (e.g. including anxiety disorders, psychoses and behavioral disorders) and across age groups, with 11 out of the 21 (52%) studies being conducted with people aged < 18 years. Although the two comparable studies (Kohn et al. 2002;Crespo, 2006) that were eligible for our review also indicate the superiority of an adapted PT compared with the non-adapted version of the same PT in a particular cultural context, both studies had methodological weaknesses and thus this issue deserves further empirical evaluation with better designed studies.
The results of the subgroup analyses found the largest effect sizes in studies of PTs using usual care/ waitlist (no treatment) control groups as compared to studies using enhanced usual care controls; and smallest effect sizes in studies using active controls (medication). This is consistent with the findings of other meta-analyses of non-adapted PTs (Peng et al. 2009;Cape et al. 2010). A considerable part of the heterogeneity in the studies in our review probably stems from the fact that the designation 'usual care' can cover a wide range of interventions, ranging from little or none to considerable. Studies that delivered 'pure PT' showed larger effect sizes than studies using PT as part of a treatment package. This finding could reflect the fact that the PT may have been less consistently delivered when it was a component of a larger package of care or complex packages may be applied to more complex patient groups where access and delivery of the intervention may pose challenges. This review is limited by the incompleteness of information for some of the studies. This occurred despite our best efforts to contact authors and search for linked papers. We attempted to look for correlations between the process and degree of adaptation ratings and effectiveness of PT by scoring each study on these dimensions and plotting this against effectiveness. No correlation was found, but this analysis was limited by the relatively small number of studies available. Furthermore, although we have reported studies that have adapted both CBT and IPT, in addition to other PTs, the small number of studies does not allow us to compare the difference in the degree of adaptation across different PTs. More complete information on the process and outcome of cultural adaptations would have been useful for drawing inferences on the method to be adopted for PT adaptation and the likely outcomes. This would have provided a stronger basis for inferring whether some of the processes involved in the adaptation were more important than others. It is also possible that publication bias inflated the estimates of efficacy for the adapted studies (just as it does in the case of non-adapted treatments).
This review extends and updates the literature on PT adaptations by including papers from non-Western countries in addition to those conducted in ethnic minority populations in Western countries, thus offering a global context to the review. Our paper also extends the existing reviews by addressing the process and outcomes of the cultural adaptations of PTs. Our findings demonstrate that cultural adaptations of PTs follow a systematic, and potentially replicable, procedure. Although the diversity of adaptations is a key finding, there were also many aspects of the treatments that did not need adaptation, indicating their universal applicability. Most adaptations were made in the method of delivery rather than in the content of the PT, emphasizing the importance of maintaining fidelity to the original treatment while allowing for greater contextual acceptability. Such adapted PTs show comparable effectiveness for the treatment of depressive disorder as reported in the larger body of evidence evaluating PTs in other populations.