Cultural adaptation of internet- and mobile-based interventions for mental disorders: a systematic review

Providing accessible and effective healthcare solutions for people living in low- and middle-income countries, migrants, and indigenous people is central to reduce the global mental health treatment gap. Internet- and mobile-based interventions (IMI) are considered scalable psychological interventions to reduce the burden of mental disorders and are culturally adapted for implementation in these target groups. In October 2020, the databases PsycInfo, MEDLINE, Embase, Cochrane Central Register of Controlled Trials, and Web of Science were systematically searched for studies that culturally adapted IMI for mental disorders. Among 9438 screened records, we identified 55 eligible articles. We extracted 17 content, methodological, and procedural components of culturally adapting IMI, aiming to consider specific situations and perspectives of the target populations. Adherence and effectiveness of the adapted IMI seemed similar to the original IMI; yet, no included study conducted a direct comparison. The presented taxonomy of cultural adaptation of IMI for mental disorders provides a basis for future studies investigating the relevance and necessity of their cultural adaptation. PROSPERO registration number: CRD42019142320.


INTRODUCTION
The World Health Organisation has called to take action in order to address global health inequalities 1 as one of the major global health challenges [2][3][4][5] . Particularly when it comes to mental health, the treatment gap in low-and middle-income countries (LMIC) is estimated to be far higher than that in high-income countries: Whereas~76-85% of people with a serious mental disorder living in LMIC do not get treated,~35-50% of people living in highincome countries do not get treated 1,6,7 . Similarly, migrants and refugees make less use of mental health services [8][9][10] , although they face many stressors before, during, and after migration 11 and show an increased prevalence of mental disorders [12][13][14] . Such a mental health treatment gap can also be observed in indigenous people, caused by lower mental health 15,16 and lower use of health services 17 . In order to improve global health equality, existing barriers of these populations need to be addressed on a structural and individual level.
Existing structural barriers, such as a lack of mental health services and difficulties in accessing services [18][19][20][21] , could be reduced by low intensity interventions [22][23][24][25] . Low intensity interventions are interventions aiming to be easy to access by a high number of people, due to a resource-saving and flexible delivery (e.g., self-help interventions, interventions delivered by nonprofessionals) 26 . Internet-and mobile-based interventions (IMI) as one kind of low intensity intervention offer anonymity, temporal and local independency, easy accessibility, and scalability [27][28][29][30] and are thus suggested to hold potential in overcoming structural healthcare barriers 31,32 . Despite reports of lower treatment completion rates 33 , IMI have been proven effective in preventing and treating mental disorders [34][35][36][37][38] . However, they were mostly developed for and evaluated in majority populations from high-income countries, and were found to be less effective for people with a differing cultural or ethnic background 39 . When it comes to the use of healthcare, populations living in LMIC, as well as migrants, refugees, or indigenous people face a range of individual barriers, such as language and cultural barriers (e.g., understanding of disease and treatment processes) or poor knowledge on the healthcare system [17][18][19][20][21]40 . These individual barriers may be reduced by considering cultural aspects in the intervention development, i.e., including knowledge on potential cultural differences 41 . However, newly developing interventions for each population requires significant effort. A more resourcesaving possibility could be to use the high number of already existing and evaluated psychological treatments and to culturally adapt them for the new target groups 42,43 .
A cultural adaptation should be conducted in a structured, systematic way by following adaptation guidelines 44,45 . Various research groups established such guidelines, focussing (a) on specific treatment components that should be culturally adapted 46-51 , or (b) on the procedure that should be followed to gradually adapt the treatment 45,[52][53][54] . The Ecological Validity Framework by Bernal et al. 47 was the first model describing in detail what to culturally adapt. The authors recommended the consideration of eight intervention components, including language (translation, differences in regional or subcultural groups), persons (patient-therapist relationship, roles), metaphors (sym-(e.g., racism). Regarding frameworks on the procedure of culturally adapting treatments, Barrera and Castro summarised various suggestions of other frameworks in their Heuristic Framework for the Cultural Adaptations of Interventions 54 . They describe the procedural components of how to culturally adapt by (1) gathering information by literature searches or qualitative research, (2) developing a preliminary adaptation based on this information, (3) testing the preliminary adaptation in case or pilot studies, and (4) refining the adaptation based on the findings of case or pilot studies.
Considering both the suggested content and procedural components of cultural adaptation intends to facilitate access to psychological treatments for people with a cultural background differing from that of the original target group 44 . Correspondingly, findings of meta-analyses aiming to investigate the relevance and necessity of cultural adaptation suggest that, in populations that the intervention was originally not developed for, culturally adapted treatments are more effective as compared to nonadapted versions of the treatment 55,56 . The higher the extent of cultural adaptation, the higher seems their effectiveness 57,58 . Based on this revealed relevance of culturally adapting face-toface treatments, it may also be of relevance to culturally adapt IMI to reach out to people with a cultural background differing from the original target group 25,59,60 . Various research groups culturally adapted IMI to fit new target groups and showed their effectiveness [61][62][63] . Furthermore, a meta-analysis indicated that an enhanced cultural adaptation of an IMI was associated with its increased effectiveness in the respective target group 59 . However, studies directly comparing the effectiveness or acceptance of culturally adapted versus non-/less-adapted IMI are lacking 59,[64][65][66] , which would allow for drawing conclusions on the added value of cultural adaptation. A precondition for such comparison trials and for examining the relevance of adapting specific components following a specific procedure would be a systematic and welldocumented adaptation 67,68 . Yet, components and procedures considered in the cultural adaptation of IMI are often only poorly reported [69][70][71][72][73] . Thus, research on the cultural adaptation of IMI is still in its early stages 66 , and there is a particular lack of knowledge on specific aspects of IMI that might need cultural adaptation in addition to the components adapted in face-to-face treatments. In this context, following aspects of evaluation frameworks of IMI could provide an orientation, such as methodological components to enhance the user engagement with the IMI, its ease of use, or design and aesthetics 74-76 . Thus, this review aims to (1) systematically identify and summarise content, methodological, and procedural components considered in the previous cultural adaptations of IMI for mental disorders; (2) summarise results on the adherence and effectiveness of culturally adapted IMI.
By providing a taxonomy on the cultural adaptation of IMI, this review may inform researchers on systematically developing scalable interventions for underserved people.

Study selection
A total of 20,012 articles were retrieved from the electronic databases; 21 articles were added by hand searching relevant articles. After removing duplicates, 9438 titles and abstracts were screened, leading to 258 records that were included in the full-text screening. In 24% of the decisions, the inclusion of a study was discussed between the reviewers. Finally, 55 studies were found eligible for the present qualitative synthesis, of which 28 provided quantitative data (see Fig. 1).

Study characteristics
All 55 included articles were published between 2012 and 2020, with 38 (69.1%) published in 2018-2020. Eight of them were study protocols, three of which additionally described conducted qualitative studies. Across the studies conducted in the included articles, a total of 3789 participants took part, plus a sample of 23,235 participants of a study analysing existing data of a patients cohort 77 . The studies were conducted or planned to be conducted in various countries: 26 in North America and Europe, 16 in Asia, 5 in Australia, 4 in the Middle East, and 2 in South or Central America. Similarly, the ethnicities of the participants were diverse across the studies: 34.6% of the studies included Asian participants, 26.9% Arabic, 15.4% European, 11.5% Northern American (including 7.7% American Indian/Alaskan Native), and 11.5% participants with other ethnic backgrounds. The included articles reported on the cultural adaptation of 42 different interventions. For 23 interventions, detailed information on their cultural adaptation had been published; for 19 interventions, less detailed information was available in published articles. Partly, more details were provided by the authors upon request (n = 9) 77-85 . Sixteen interventions described in the articles addressed depression, also in combination with anxiety, ten addressed stress or traumatisation, four substance abuse disorders, and the remaining interventions addressed other mental disorders or a combination of the named ones.
An overview of the study characteristics can be found in Table 1. Details on the interventions used in the studies are illustrated in Supplementary Table 3.

Quality assessment
The initial interrater reliability between the two independent raters was κ = 0.62, which can be considered a substantial agreement 86 . Overall, the quality of the included articles revealed an average quality score of 75.4%. The assessed quality score ranged from 35.7 (ref. 87 ) to 95.2% (ref. 63 ). Nearly all articles provided an explicit theoretical framework (93.9% of the maximum score), named the aims of their conducted or planned study (95.8%), and described their research setting (93.9%) and their procedure for data collection (90.3%) in a satisfactory way. Often, there was only poor evidence of the sample size (40.1%), and users were rarely involved in the study design (42.4%). Table 1 presents the individual quality score of every article and Supplementary Table 4 contains the comprehensive ratings of the two independent reviewers.

Components of cultural adaptation
Various components of which the authors reported the cultural adaptation were identified across the included articles and grouped in distinguishable aspects of culturally adapting IMI. Using the frameworks of previous literature as a base 47,54,[74][75][76] , the classification had been continuously refined in the course of data synthesis, resulting in ten content, four methodological, and three procedural components that provided qualitative information on the cultural adaptation of IMI for mental disorders. Generally, the consideration of these components was used by the authors to enhance the fit of the IMI to the context, lives, and burdens of the new target group. Across all interventions averaged consideration of content (68.8%) versus methodological (69.0%) versus procedural (66.7%) components was comparable with 68.5% of all components being considered. The 17 components of culturally adapting IMI that were identified across the articles are illustrated in Table 2, including specific adaptation examples of the articles. A more detailed illustration of the conducted adaptations of each IMI can be found in Supplementary Table 5. An overview of the 17 components illustrating the respective extent of consideration across the articles is provided in Fig. 2.
Regarding content components, (1) an important element to adapt were the characters illustrated in the IMI (adapted in 35 of 42 interventions). (2) Alongside the characters, activities were added, removed, or tailored to make them relevant to the target group (25/42 IMI). (3) Furthermore, environments, settings, and, associated with this, specific burdens illustrated in the IMI were adapted (36/42 IMI). (4) Related to these aspects, researchers changed the content to accommodate culturally relevant values and traditions (28/42 IMI). Beside the adaptation of the daily life illustrated in the IMI, language adaptation was considered important: (5) numerous studies translated the entire intervention into the main language or the regional dialect of the target group (35/42 IMI). In addition, language and texts were (6) tailored to the target group (34/42 IMI), and (7) visualised with appropriate quotes, symbols, and metaphors (18/42 IMI). (8) When culturally adapting their IMI, researchers also considered potential differences in the target groups' concepts of mental health and its treatment (26/42 IMI). Correspondingly, (9) the aims (23/42 IMI) and (10) the used treatment methods in the IMI (27/42 IMI) were changed occasionally.
Regarding methodological components of cultural adaptation, (11) the general structure of IMI was adapted (34/42 IMI). (12) Also, researchers adapted their IMI to enhance its functionality and ease of use for the specific target group (26/42 IMI), e.g., by including interactive elements or simplifying the navigation. (13) Moreover, the design and the aesthetics were adjusted (34/42 IMI). (14) As another aspect of culturally adapting IMI, researchers have changed the format or amount of guidance (22/42 IMI).
With regard to the procedure of cultural adaptation, (15) researchers used a wide range of methods to obtain information about necessary cultural adaptations of their IMI and about the adequacy of the conducted adaptations (32/42 IMI used at least two different methods). (16) During the cultural adaptation process, different people were involved, including healthcare professionals or the target group itself (32/42 IMI involved at least two groups of people). (17) Many authors also based their adaptation process on theoretical frameworks for cultural adaptations of face-to-face therapies (20/42 IMI).

Extent of cultural adaptation
The consideration of each of the 17 components of cultural adaptation of IMI was evaluated across the articles to illustrate the extent of cultural adaptation for each intervention. The points were added up (0-17 points), revealing an average extent of cultural adaptation of M = 11.6 (SD = 3.7), with the smallest extent being three (17.7%) 88 and the highest being 17 points (100%) 69,89-91 for the respective interventions. A list of articles considering the respective components is provided in Table 3.
Adherence and effectiveness of the culturally adapted interventions Characteristics of those articles providing quantitative data, and their adherence and effectiveness outcomes are illustrated in Table 4, including the measures of the respective non-adapted interventions, if available.
Eight single group studies reported on the effectiveness of their IMI comparing mental disorder symptoms of the participants between pre-and post-treatment assessments, three of which assessed the effectiveness of the same intervention [97][98][99] . Fourteen (randomised) controlled trials (Chen et al. 81 stopped randomisation) reported on the effectiveness of their IMI comparing mental disorder symptoms of the participants who had access to an adapted IMI to those who had no access (control group receiving delayed access, no access, or psychoeducational information). Outcomes of one randomised controlled trial were published in two articles 100,101 . Primary outcomes varied between the studies, with the Patient Health Questionnaire 102 being the scale most used (n = 9), followed by the General Anxiety Disorder Scale 103 (n = 4) and the Post-traumatic Disorder Scale 104 (n = 3). Similarly, the time between pre-and post-assessments differed between the studies, ranging from 4 weeks 96 to 6 months 105 . Effect sizes of randomised controlled trials ranged from Hedges' g = 0.15 (ref. 100 ) to 2.70 (ref. 106 ; see Fig. 3).

Post hoc analyses
Correlation analyses did not reveal any significant links between the intervention characteristics (amount of modules, provided guidance) and the extent of (specific components of) cultural adaptation, r(40) ≤ 0.12, p ≥ 0.330. The extent of culturally adapting content components significantly differed between the target groups of the IMI (people living in LMIC, in high-income countries, immigrants, indigenous people), F(3,38) = 6.09, p = 0.002, with pairwise comparisons suggesting a lower extent of content cultural adaptation for people from high-income countries, as compared to migrants and people living in LMIC. No link was revealed between the amount of modules, guidance, or extent of cultural adaptation and the effectiveness (r(13) ≤ 0.20, p ≥ 0.487) or adherence of the IMI (r(9) ≤ −0.35, p ≥ 0.287). Nor did the

DISCUSSION
This is a systematic review synthesising the components and procedures used in the cultural adaptation of IMI for mental disorders. The proposed taxonomy comprises 17 components, covering the key domains of content (e.g., illustrated daily life and mental health concepts), methodological (e.g., structure and design), and procedural components (e.g., used methods and involved persons). The 55 articles included in this review explored, conducted, and reported on the cultural adaptation of 42 IMI that were originally used for a culturally differing target group. Cultural adaptations were done to illustrate relatable situations (e.g., persons, activities), to fit the characteristics of the communities (e.g., collectivism, values, traditions), to consider specific risk factors for low mental health (e.g., burdens due to migration, war, discrimination, and  Fig. 2 Taxonomy of cultural adaption of internet-and mobilebased interventions (IMI). Percentages of the 42 included IMI that fulfilled the respective content, methodological, and procedural components of the taxonomy.   low socio-economic status), and to address health-related problems (e.g., low mental health literacy and corresponding distrust, risky behaviour, or limited access to treatment). Twenty-eight studies, including 14 (randomised) controlled trials, evaluated the culturally adapted IMI in terms of adherence and/or effectiveness for the new target group. Adherence rates and effect sizes of randomised controlled trials seem to be comparable to both those found in the studies on the respective original IMI (see Table 4), and those found in studies investigating the adherence and effectiveness of IMI in general 36,107 . However, enrolment and adherence rates of IMI are generally low [108][109][110] . Above that, we could not identify any study on a direct comparison of a culturally adapted with a non-adapted IMI. Hence, we believe it is premature to conclude that cultural adaptation is a prerequisite to make IMI work in people living in LMIC, migrants, or indigenous people. Our taxonomy of cultural adaptation of IMI may provide a base for future researchers to systematically adapt IMI by considering the 17 identified components. The included ten content components overlap with components that are culturally adapted in face-to-face psychological treatments 47 . Thus, both surface (e.g., the adaptation of characters, activities, and language) and deep (e.g., the adaptation of burdens, values, mental health concepts, and goals and methods of treatment) structure adaptations were shown to be important 50 . Also, the three procedural components seem to reflect the procedure proposed in the adaptation of faceto-face psychological treatments 54 . Accordingly, the study investigators used multiple methods to obtain information from various groups of people (e.g., mental health experts, target group) on necessary adaptations of the IMI, including focus groups and feasibility trials. Thereby, about half of the IMI were adapted following theoretical frameworks of cultural adaptation of face-toface treatments. However, in addition to the content and procedural components, we identified four unique methodological components specifically considered in the cultural adaptation of IMI. These include components related to the structure (e.g.,  Knaevelsrud (2015) Lin (2020) Lindegaard (2020) Luo (2020) Nygren (2019) Paris (2018) Salamanca-Sanabria (2020) Silva (2020) Titov (2018) Ünlü Ince (2013) Wagner (2012) Wang (2013) Hedges' g Culturally adapted study Original study Fig. 3 Effect sizes (Hedges' g) of the culturally adapted versus original internet-and mobile-based interventions. Twenty-two included articles provided relevant information on the effect of their intervention. If more than one primary outcome was reported, the effect size of the Patient Health Questionnaire 102 is illustrated. Both single group studies and randomised controlled trials are included. Full information on the respective effect sizes can be found in Table 4.
shortened text sequences or modules), functionality (e.g., the consideration of a potentially bad internet connection), design and aesthetics (e.g., the amount or design of pictures), and human guidance (e.g., level of guidance). Hence, it may not be sufficient to rely on existing frameworks for cultural adaptation of face-toface treatments when culturally adapting IMI, as also highlighted by Lal et al. 72 and implemented by Burchert et al. 111 .
By enabling a systematic cultural adaptation of IMI, our taxonomy on cultural adaptation of IMI can furthermore serve as a base to measure the extent and type of cultural adaptation, and thus can lay an initial foundation for further examining the relevance and necessity of the cultural adaptation of (specific elements of) IMI. This relevance and necessity of cultural adaptation could differ, for example, depending on the target group, the addressed disorder, or the language of the IMI. Conducted exploratory post hoc analyses indicate that culturally adapting IMI for people with a migrant background and people living in LMIC might be more important as compared to culturally adapting IMI for new target groups in western countries, especially with respect to content components. However, in the current systematic review, no link between the extent of the conducted cultural adaptations, and the effectiveness or adherence of the IMI was revealed. The analyses should be interpreted with caution, as the present systematic review included very heterogeneous IMI and did not yield sufficient outcome data. Even with more randomised controlled trials included in meta-analyses, the heterogeneity of interventions studied remains mostly very high, making it difficult to draw conclusions on the influence of specific intervention characteristics or components of cultural adaptation on the effectiveness of the IMI. Thus, questions on possible differences in the necessity of cultural adaptation and its influence on the effectiveness of IMI could be addressed in other study designs. By using randomised factorial trials 112,113 , different versions of the same IMI that were systematically culturally adapted to varying degrees could be tested against each other. For example, deep structure content components versus surface structure content components versus no content components of an IMI could be culturally adapted. The effectiveness of each of the versions could then be evaluated and compared, herewith enabling to draw conclusions on the benefit of culturally adapting these specific components.
Some limitations of this systematic review need to be considered. First, we excluded studies that did not report (n = 14) or provide details on (n = 6) their cultural adaptation. At the same time, we excluded grey literature with potential unpublished reports on conducted cultural adaptation. This might have led to biased findings and hindered us from drawing conclusions on a potential influence of the (extent of) cultural adaptation of IMI on the adherence or effectiveness of the respective IMI. Second, we also excluded studies that newly developed culturally sensitive IMI for a target group of people living in LMIC, migrants, or indigenous people (n = 32), in contrast to studies that culturally adapted previously used and evaluated IMI. Yet, the 17 components of the taxonomy of cultural adaptation might also be helpful in the development of culturally sensitive IMI. Thus, it may be a fruitful next step to conduct a review on the development of culturally sensitive IMI and compare the results with the current review. Third, the wide range of target groups in our review makes it difficult to interpret the findings across all included target groups. Following the cultural diversity approach 114 , the necessary extent and depth of culturally adapting IMI may differ between target groups, also depending on the similarities and differences between the original and new target groups. Fourth, although we extracted and summarised aspects of cultural adaptation that the authors have reported to be adapted in order to enhance the cultural fit, we cannot fully rule out that there might have been other reasons (e.g., resources of the researchers and technical innovations). Fifth, the choice of the two cultural adaptation models used as a basis for data extraction 47,54 was not based on a systematic approach, but mainly on the frequent use of the models in previous research. Sixth, as the extracted components of cultural adaptation represent differing intensities of adaptation (surface structure adaptations, such as language translation, versus deep structure adaptations, such as the consideration of differences in the concepts of mental health and its treatment), the extent to which the components are considered is not equal to the extensiveness of cultural adaptation. Seventh, reliability regarding data extraction and data synthesis was not evaluated and may thus be limited.

CONCLUSIONS
The mental health treatment gap for people in LMIC and for migrants or indigenous people in high-income countries needs to be addressed with innovative and scalable approaches 5 . To provide impactful interventions, an efficient and systematic cultural adaptation is central 24 . The present systematic review illustrates a taxonomy of 17 components of culturally adapting IMI, making them more relatable to the new target group by considering their specific context, burdens, and understanding of mental health. This taxonomy can serve as a base for the systematic cultural adaptation of IMI. The next steps should be to investigate the relevance and necessity of a systematic cultural adaptation of IMI, which could help provide adequate and effective mental healthcare to underserved populations, herewith contributing to reduce global public mental health inequalities.

METHODS
This systematic review is reported according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses, the PRISMA guidelines (see Supplementary Table 1 for PRISMA checklist) 115 . A systematic review protocol has been published 116 and registered in the International Prospective Register of Systematic Reviews of the National Institute for Health Research (PROSPERO; registration number: CRD42019142320).

Eligibility criteria
We included (a) peer-reviewed journal articles that reported on an intervention that (b) was primarily provided in an internet-, computer-, app-, or mobile-based setting (with or without human support; pure videoconferencing interventions were excluded), (c) used psychological methods to address mental disorders or disturbances, and (d) was adapted from a previously used and evaluated IMI to fit a new and culturally differing target group. Interventions newly developed for a culturally differing target group were excluded due to assumed differences in the procedure and content of cultural adaptation versus culturally sensitive development of IMI. Psychological interventions were defined according to Kampling et al. and include, among other psychological-oriented interventions, cognitive behaviour therapy, psychodynamic psychotherapy, behaviour therapy or behaviour modification, systemic therapy, third-wave cognitive behaviour therapies, humanistic therapies, and integrative therapies 117 .
There were no restrictions in the participants' age nor the language of the articles. Articles were translated to English, if necessary. Furthermore, study protocols were included, if they provided information on the cultural adaptation of the used interventions.
Search strategy A systematic literature search was conducted on October, 15th 2020 in EbscoHost/Medline, EbscoHost/PsycINFO, Ovid/Embase, Cochrane Central Register of Controlled Trials (CENTRAL), and Web of Science. The search covered two parallel reviews, with the second review including articles on the cultural adaptation of IMI for health promotion 118 . Keywords of the search term for the present review included a combination of terms and synonyms of (1) internet-based interventions, (2) mental health, and (3) cultural adaptation. The search was complemented by hand search, as well as backward and forward searches. The publication of studies described in an included study protocol was checked before the final data extraction. The full search term can be found in Supplementary Table 2.

Study selection
The selection of articles was conducted in two steps using Covidence 119 . First, two independent reviewers screened titles and abstracts of all articles against the inclusion criteria. Second, the full-texts of seemingly relevant or ambiguous articles were screened against the inclusion criteria by two independent reviewers. Reasons for exclusion were documented. Disagreement was resolved in discussion, and, if no consensus could be reached, by consulting a third reviewer. The study selection process is illustrated in the PRISMA flow chart (see Fig. 1).

Data extraction
Using a data extraction form in Covidence 119 , the following information was extracted from the selected articles: (a) study identification items, (b) sample characteristic, (c) study design characteristics, (d) study settings, (e) original intervention characteristics, (f) culturally adapted intervention characteristics, (g) details on content and methodological components of cultural adaptation, (h) details on procedural components of cultural adaptation, as well as (h) primary and secondary outcomes of the studies (adherence and effectiveness). The study authors were contacted if the variables of interest were not available or not clearly presented in the studies. All extracted data was doublechecked by an independent reviewer.

Quality assessment
Two independent reviewers assessed the Quality Assessment Tool for Reviewing Studies with Diverse Designs (QATSDD) 120 . The tool comprises 16 criteria that are applicable to diverse research designs, of which 14 apply to quantitative and to qualitative methods respectively. Thus, the QATSDD enabled an overall quality rating for included quantitative, qualitative, and mixedmethods studies. Beside items assessing completeness and clarity of reporting, further items can help rate the risk of bias of the studies (e.g., fit between stated research question and method of data collection, format and content of data collection tool, or method of analysis; good justification for analytical method selected). The criteria were rated on a four-point scale, producing a global score showing a low or high study quality. The quality score is presented as percentage of the highest possible points. Arising discrepancies between the two reviewers were resolved in discussion; if further needed, a third reviewer was consulted. Cohen's Kappa was used to assess interrater agreement between low (0 and 1 point) and high (2 and 3 points) study quality 121 .

Data synthesis
A qualitative synthesis of all included articles was conducted to describe the characteristics and findings of the studies. To answer the research questions, details on the cultural adaptation reported in the included studies were summarised and classified; only adaptations that the authors reported to have been made for cultural reasons were considered. The classification included content components, based on Bernal et al. 47 , methodological components, based on IMI evaluation frameworks [74][75][76] , and procedural components, based on Barrera and Castro 54 . We additionally rated which of the extracted components were culturally adapted versus not adapted in which intervention (one versus zero point), so to provide descriptive cultural adaptation scores to illustrate the extent of the conducted cultural adaptation for each of the interventions, and each of the components.
Findings related to the adherence to and the effectiveness of the culturally adapted IMI for mental disorders were summarised and descriptively illustrated. Adherence was evaluated using the percentage of completed modules (average number of completed modules divided by the number of all modules of the IMI). In addition, the percentage of study participants who completed all modules of the IMI was illustrated. The effectiveness of the intervention was evaluated using the effect size Hedges' g 122 . In case of conducted randomised controlled trials, the effect size was calculated for the mean differences in the primary outcome(s) of the intervention and control groups at post-treatment assessments. In case of single group study designs, the effect size was calculated for the mean differences in the primary outcome(s) between pre-and post-treatment assessments. If available, adherence and effectiveness outcomes of the original IMI were illustrated in comparison to those of the culturally adapted IMI.

Post hoc analyses
To exploratively investigate intervention characteristics that might influence the conducted cultural adaptation and the effectiveness and adherence of the adapted IMI, we conducted post hoc analyses. First, potential links between intervention characteristics (duration, provided guidance, and target group of the IMI) and the extent of cultural adaptation or the extent of specific components of cultural adaptation (content, methodological, and procedural) were analysed using Pearson's correlation coefficients and an analysis of variance. Second, potential links between intervention characteristics or the extent of cultural adaptation and the effectiveness or the adherence (percentage of completers) of the interventions were analysed also using Pearson's correlation coefficients or analyses of variance. Due to the explorative character of the analyses, no adjustments for multiple comparisons were applied.

DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable request.