Implementation of a task-shared psychosocial intervention for perinatal depression in South Africa: A grounded theory process evaluation

Abstract

Perinatal depression carries significant levels of disability for both women and their infants, but there is a large treatment gap for this condition in low and middle-income countries (LMICs). To address this gap, task-sharing using community health workers (CHWs) to provide psychosocial interventions for perinatal depression is increasingly common. Many of these interventions have shown significant positive effects on maternal mood, however not all have done so. This study used data from a task-shared psychosocial intervention for perinatal depression in Cape Town, South Africa, the ‘Africa Focus on Intervention Research for Mental Health’ (AFFIRM-SA) randomised controlled trial (RCT). It aimed to examine the processes that occurred within the delivery of the counselling intervention, and to use these findings to provide recommendations for psychosocial task sharing in LMICs.

A grounded theory analysis was conducted of 234 counselling session transcripts from 39 randomly selected participants from the RCT. This revealed that the effectiveness of the intervention was compromised by the negative influence of participants’ socio-economic context, and by counselling strategies that did not align with what was intended in the counselling manual. Despite this, participants provided spontaneous accounts of improvement in mood and cognition, and reasons for these improvements, interpreted as elements that were therapeutically effective for them. Most of these elements aligned with previously identified ‘common elements’ of therapy.

Recommendations for future research and practice include conducting participatory formative research, using an iterative and responsive research design informed by implementation science, incorporating contextually appropriate strategies in interventions such as addressing social determinants of mental health, conducting intensive training and supervision, adopting a staged approach to managing depression, and using common elements of therapy as the basis for psychosocial interventions.

1. Introduction

Women in the perinatal period are vulnerable to depression. This creates significant levels of disability for women and their infants, including preterm delivery and low birth weight (Hollins, 2007; Stewart, 2011), poor mother-infant relationships and child growth (Cooper et al., 1999), and impaired behavioural and cognitive development in children (Field et al., 2006; Surkan et al., 2011). The prevalence of perinatal depression is particularly high in low- and middle-income countries (LMICs), where estimates have shown a pooled prevalence of antenatal depression at 19.2%, and 18.7% postnatally (Woody et al., 2017). This is almost double that of high-income countries (Woody et al., 2017).

Multiple factors play a role in increasing risk for perinatal depression in women in LMIC. These include having low education levels (Coast et al., 2012), food insecurity (Lund et al., 2010; Tsai et al., 2012), being the victim of abuse (Dadi et al., 2020; Mitchell et al., 2016), having an unplanned or unwanted pregnancy (Fisher et al., 2012; Hartley et al., 2011), lack of social support (Dadi et al., 2020; Smyth et al., 2015), HIV (Berger-Greenstein et al., 2007; Kagee & Martin, 2010), and other conditions of poverty such as low or irregular income levels or low socio-economic status, inadequate housing, overcrowding and lack of privacy, high levels of crime and violence, high prevalence of communicable diseases, and poor sanitation and access to basic services (Dadi et al., 2020; Gelaye et al., 2016; Patel et al., 2010).

Despite the high prevalence of perinatal depression and its implications for infants, it largely does not receive the treatment and service personnel it deserves (Gelaye et al., 2016), creating what is referred to as a treatment gap (Saxena et al., 2007). Task sharing, involving non-specialist health care workers delivering evidence-based mental health treatment through routine delivery systems, has been recommended to increase coverage of care for mental disorders at primary care levels and to address this gap in LMICs (WHO, 2008).

Testing of task-shared interventions and treatments for a variety of mental illnesses in LMICs using randomised controlled trials (RCTs) over the last two decades has contributed valuable evidence to the field of global mental health, with many of these interventions demonstrating significant beneficial effects (Barbui et al., 2020; Singla et al., 2017). Despite this, the transmission of research into policy for treatment at primary health care facilities is still slow.

For perinatal depression, task-shared interventions have similarly shown significant benefits (Clarke et al., 2013; Rahman et al., 2013). Factors that contributed to these impacts included the relevance of the therapeutic relationship between CHW and participant, a sense of trust, being non-judgmental and empathic, the use of local CHWs who understood their clients’ sociocultural backgrounds, and the importance of adapting interventions to the circumstances that women live in (Rahman et al., 2013).

However, more recent iterations of evidence-based interventions have demonstrated only small improvements in perinatal depression, such as those in India, Nigeria, Pakistan and South Africa (Fuhr et al., 2019; Gureje et al., 2019; Lund et al., 2020; Sikander et al., 2019). Although these trials have not yet published reviews on their implementation processes, some hypothesized reasons for the findings are that all of these interventions had enhanced usual care in their control groups, which may have had a therapeutic effect itself, and that by the six month outcome points, there may have been spontaneous remission of depression in both groups. To the best of our knowledge, none reviewed what had happened during the psychotherapy sessions to try and understand why the process worked or not.

This mixed evidence demonstrates that task sharing is not a silver bullet for the provision of mental health services in LMICs (Padmanathan & De Silva, 2013); although it is currently the most viable approach to improving access to treatment for perinatal depression and other common mental disorders (CMDs) (Bolton, 2019). Evaluations are therefore needed to examine exactly how task sharing treatments work or don’t work, and what happens in the implementation process, so that they can more adequately address the needs of underserved communities (Barnett et al., 2018; Cuijpers, 2019), and contribute to reducing the treatment gap in similar low-income settings.

Implementation science is an approach that assists in improving quality, effectiveness, and sustainability of health services (Curran et al., 2012; Eccles & Mittman, 2006). This type of research incorporates the assessment of progress and barriers and facilitators during implementation in order to identify and rectify these if needed (Damschroder et al., 2009).

Process evaluations are another type of approach that complement RCTs through the assessment of mechanisms of change that impact the effectiveness of interventions (Cuijpers et al., 2019; Singla et al., 2017). Moore et al. (Moore et al., 2015, p. 350) describe the function of process evaluations to gain understandings of the pathways that link the theory underlying an intervention and its causal assumptions, to the outcomes produced. This can be done through examining contextual factors, implementation factors, mechanisms of impact, and the links between these.

The few studies that have conducted process evaluations of task-shared interventions for CMDs in LMICs have identified the mechanisms that led to effective treatment in the interventions (Jordans et al., 2012; Petersen et al., 2012), and the barriers and facilitators to session attendance, uptake of practices and content, and training and supervision (Munodawafa et al., 2017; Selohilwe et al., 2019; Singla & Kumbakumba, 2015).

Despite these evaluations, evidence reporting the implementation processes of mental health programmes in LMICs is still scarce (Esponda et al., 2020). This study aimed to conduct a process evaluation of a task-shared psychosocial intervention for perinatal depression in South Africa through a grounded theory analysis. More specifically, it aimed to identify processes that occurred within the delivery of counselling sessions in the intervention, and to use these findings to share recommendations for the development of future task-shared interventions for CMDs in LMICs.

2. Methods

2.1. The parent study: The AFFIRM-SA RCT

The study used data from the ‘Africa Focus on Intervention Research for Mental Health - South Africa (AFFIRM-SA)’ RCT, that was conducted between 2011 and 2016. A full description of the RCT sampling, recruitment and intervention methods are described in Lund et al. (Lund et al., 2014, 2020). In brief, the RCT aimed to test a six-session task--shared manualised counselling intervention delivered by community health workers (CHWs) to depressed pregnant women at two Midwife and Obstetrics Units (MOUs) in Khayelitsha, Cape Town, South Africa. Khayelitsha is a peri-urban township on the outskirts of Cape Town, characteristic of many LMIC urban settlements, with poor health service provision and high levels of poverty and crime (Statistics South Africa, 2011). Formative research was conducted before implementation of the trial through interviews with midwives, HIV counsellors, CHWs, and depressed and non-depressed pregnant women (Davies et al., 2016). This aimed to gain understanding of local isiXhosa women’s conceptions of depression and its causes, how this related to international classification systems for depressive symptoms, and perspectives on practical elements of implementing counselling at MOUs in Khayelitsha. This informed the development of the intervention content. A local functional assessment instrument was also developed to assess potential impact of improvement in depressive symptoms on functioning in a manner that was relevant to activities of daily living by perinatal isiXhosa-speaking women in Khayelitsha (Schneider et al., 2015). This was in keeping with previous approaches to develop culturally valid functioning assessment instruments (Bolton, 2019).

Following this, a counselling manual was developed through review of manuals from previous task-shared single-modality interventions in LMICs, and consultation with an expert panel (Nyatsanza et al., 2016). Various evidence-based modalities were recommended by the panel, including Cognitive Behavioural Therapy (CBT), Behaviour Activation, Problem solving and Psychoeducation. It was hypothesized that these modalities would improve perinatal depression through the mechanisms associated with each, such as reframing negative cognitions around individual circumstances through CBT, engaging in activities to improve well-being and social support through behavioural activation (Singla et al., 2021), and identifying and prioritising problems and potential solutions with problem solving. The panel decided to combine elements of all of these modalities in an eclectic approach. The intervention was then piloted with six participants, and feedback was obtained from the counsellors regarding practical implementation issues.

Following informed consent, participants were recruited into the main trial if they were 18 years or over, were pregnant, no later than 28 weeks gestation, spoke isiXhosa, and screened positive for depression with a cut-off of 13 or above on the Edinburgh Postnatal Depression Scale (EPDS) (Cox et al., 1987), validated in Khayelitsha and Johannesburg in South Africa (De Bruin et al., 2004; Lawrie et al., 1998). The 17-item Hamilton Depression Rating Scale (HDRS) (Hamilton, 1986), was used as the primary outcome measure. This was translated, adapted and validated for use in isiXhosa by non-clinicians (Davies et al., 2019a). Ethical approval was received from the University of Cape Town Human Research Ethics Committee (Ref. 226/2011) and the National Institute of Mental Health (NIMH).

Intervention counsellors were recruited from a local non-governmental organisation (NGO) in Khayelitsha where they were employed to provide informational health visits to pregnant women and mothers of young babies in their communities. Twelve CHWs identified by the NGO were trained over five days to implement the six-session manualised counselling intervention for women allocated to the intervention arm in the study. Six CHWs were then selected on the basis of their performance during the training, including their understanding of the material, their level of interest, and motivation. Pregnant participants attended sessions on average every two weeks. Counsellors aimed to complete all sessions before women gave birth, but this was not always possible due to varying participant circumstances impacting their ability to attend regular sessions.

Sessions covered the following therapeutic modalities and topics: 1) psychoeducation, 2) problem solving, 3) behavioural activation, 4) healthy thinking, 5) birth preparation, and 6) termination and evaluation. Every session started with an introduction and summary of the previous session, and then an explanation of the session topic, which included in-session activities for participants to complete. Homework on the topic was assigned, and counsellors summarised the session before concluding it. Control arm participants received three monthly phone calls involving standardised health questions from separate CHWs not trained in counselling. All counselling sessions were audio recorded with consent. A clinical social worker employed as counselling supervisor listened to samples of the recordings and conducted weekly supervision with the counsellors. If women reported current partner abuse or severe psychological distress, they were referred to the social worker or mental health nurse at the clinic and retained in the trial. Follow up assessments were conducted at one month prior to birth, and at three- and 12-months postpartum.

A total of 425 participants were recruited into the trial and 209 were randomly allocated to the intervention arm, with 97 (46%) of these completing all six sessions of counselling. Of the 216 allocated to the control arm, 187 (87%) received all three phone calls. As reported in Lund et al. (Lund et al., 2020), there was no significant difference in response on the HDRS between intervention and control arms at three months postpartum (Risk Ratio (RR): 1.11, 95% Confidence Interval (CI): 0.87–1.43, p = 0.392). There was however a significant improvement in EPDS scores in the intervention arm (mean = 7.6, SD: 5.2) compared to the control arm (mean = 9.5, SD: 5.7; RR = 0.78, 95% CI: 0.67–0.91, p = 0.001), which was maintained at 12 months postpartum.

2.2. Current study

The current study used data from participants who had been in the intervention arm of the RCT, once the trial had been completed. Ethical approval was obtained from the University of Cape Town to use these data in this analysis (Ref. 790/2015). From the 97 participants who completed all six counselling sessions, seven participants per counsellor (N = 42) were randomly selected for transcription of the audio recordings of the sessions. Three of these participants had incomplete transcriptions due to problems with the audio recordings identified by the transcriber, amounting to an analysis of 234 transcribed sessions from 39 participants (six transcribed sessions per participant).

Table 1 below provides a comparison of data from the study sample (n = 39), those not selected for the sample (n = 58), all participants who completed sessions (n = 97), the intervention arm full sample (n = 209), and the 425 participants who completed the baseline assessment for the RCT as a whole. Scores are very similar for the baseline characteristics across samples, other than that the randomly selected sample for this grounded theory analysis had a higher percentage of women who were employed at baseline compared to those not selected for the sample. This difference is not significant (p = 0.146), and the sample is otherwise representative of the 97 participants who completed the intervention and the 209 intervention arm participants.

Table 1.

Participant characteristics.

Characteristics	Study sample(n = 39)	Not selected for this study (n = 58)	P value (sample vs not selected)a	Completed all sessions (n = 97)	RCT intervention arm sample (n = 209)	RCT total sample (n = 425)
Mean Age (SD)	27.7 (5.97)	27.9 (5.59)	0.977	27.8 (5.71)	27.59 (5.94)	27.31 (5.71)
School level completed*(Mean,SD)	10.9 (1.47)	11.0 (1.28)	0.702	10.9 (1.35)	10.92 (1.65)	11.00 (1.43)
Percentage employed (n, %)	22 (56.41)	23 (39.66)	0.146b	45 (46.39)	98 (46.89)	193 (45.41)
EPDS score Baseline (Mean, SD)	17.2 (2.92)	18.0 (3.75)	0.371	17.7 (3.45)	17.57 (3.73)	17.36 (3.64)
HDRS score baseline (Mean, SD)	15.9 (5.27)	14.8 (4.33)	0.426	15.2 (4.74)	15.53 (4.88)	15.49 (4.76)
EPDS score 3 Months postnatal (Mean, SD)	7.7 (5.45)	7.2 (4.98)	0.699	7.4 (5.15)	7.61 (5.18)	8.65 (5.55)
HDRS score 3 Months postnatal (Mean, SD)	8.5 (4.63)	8.6 (3.77)	0.712	8.5 (4.10)	9.11 (4.56)	9.68 (4.81)

^*South African schooling grades are from grade 0 to 12.

^aMann-Whitney U test.

^bFisher’s Exact test.

Table 2 presents basic characteristics of the six counsellors in the study. Participants were anonymised using successive alphabetical lettering corresponding with their counsellor’s codes (e.g. Counsellor A (Csr A), participant AA, AB etc.). Session numbers were abbreviated as S1–6.

Table 2.

Counsellor characteristics.

Counsellor	Age	Marital Status	School level completed	Years providing health visits to pregnant women before AFFIRM-SA training
Csr A	46	Widow	Grade 9	2.5 years
Csr B	44	Single	Grade 12	5 years
Csr C	40	Widow	Grade 12	11 years
Csr D	32	Single	Grade 11	2 years
Csr E	28	Married	Grade 12	4 years
Csr F	33	Single	Grade 12	2 years

Grounded theory (Corbin & Strauss, 2014) was employed to analyse the counselling session transcripts. This approach uses a method of ‘constant comparative analysis’ to develop an ‘emergent grounded theory’ that evolves as the analysis is conducted (Strauss & Corbin, 1998). This helped to avoid bias by not imposing preconceived theories or ideas of what occurred during the counselling sessions, onto the data. The analysis involved a process of ‘open’, ‘axial’, and ‘selective’ coding of the transcripts (Strauss & Corbin, 1998). Open coding comprised reading all of the transcripts and coding anything that was of interest in the interactions between counsellor and participant that indicated processes that were occurring within the counselling sessions. ‘Axial coding’ then involved identifying sub-categories and relationships between the ‘open codes’. ‘Selective coding’ was done by identifying some key themes that might explain the reasons for the various codes identified. Theoretical and data saturation was noticeably reached from the available transcripts (Fusch & Ness, 2015).

This approach to analysis facilitated a deeper understanding of the nuances of CHWs implementing psychosocial counselling, and allowed for lessons and recommendations to be drawn from the data. To the best of our knowledge, this is the first analysis of its kind from a task shared counselling intervention in an LMIC.

The analysis was guided by Murdoch’s framework for process evaluations for complex health interventions (Murdoch, 2016). This framework is guided by the concepts of ‘linguistic ethnology’ (Rampton et al., 2004) and ‘text trajectories’ (Woydack & Rampton, 2016), and draws on Harris and Rampton’s (Harris & Rampton, 2009) four levels of context that influence the dynamic relationship between structure and action in intervention delivery. These include macro, meso, micro linguistic and micro non-linguistic contextual features. This framework is helpful in that it proposes an examination of the enactment of protocols, through identifying whether the intervention was delivered as intended, and was congruent with the initial hypothesis of the trial. In addition, it directs attention to the various levels of context that influence the impact of an intervention, from the initial development of the theory of an RCT through to the eventual receiving of it (Murdoch, 2019). The limitations of using the framework include that the analysis only examines a sample of texts, which may miss “a particular configuration of contextual features” (Murdoch, 2016). However, the process still encourages a scrutiny of trial implementation through the lens of an intervention inevitably being produced by and through its particular context (Murdoch, 2016).

Data analysis was conducted by the first author, who was a Research Officer in the RCT, responsible for training and supervising the field-workers who collected data for the RCT. She was not involved in the development or design of the intervention or control group protocols. Since two raters were not used, the themes, codes, and coded data were then presented to a specialist panel for peer examination (Krefting, 1991; Lincoln & Guba, 1985), to assess acceptability, trustworthiness and rigour of the analysis and coding (Barber & Walczak, 2009). The panel comprised the RCT Principal Investigator and Project Manager, who are the co-authors of this paper, and two psychological counsellors with experience in low-income settings, not involved in the RCT. All of the panel had conducted qualitative data analysis and worked in implementation of task-shared psychological interventions. The members of the panel were given advanced preparation for the presentation, were able to review the data, and met face-to-face, as recommended by Barber and Walczak (Barber & Walczak, 2009). The panel endorsed the acceptability and trustworthiness of the coding and of the distinctions between codes, as well as the rationale for the way the data had been coded.

3. Results

Three broad categories emerged from the grounded theory coding that represented some of the processes that occurred during the counselling sessions, and which had varying influences on intervention delivery and effectiveness. These were: the influence of context on therapeutic effectiveness, counselling factors affecting quality of delivery, and elements perceived by participants to be therapeutically effective.

3.1. Influence of context on therapeutic effectiveness

The first broad category identified was that of participants’ socio-economic context and its potential influence on the therapeutic effectiveness of the sessions. The data demonstrated that all of the participants faced persistent poverty in their lives, which interacted in a negative cycle with factors such as food insecurity, lack of support from partners, physical, sexual and economic abuse, alcohol abuse by partners, HIV/AIDS and continuously occurring traumatic events. These factors impeded potential efficacy of the sessions by ‘interrupting’ the improvements in mood or reduction in symptoms through their consistent presence in the women’s lives, and conversely, through the sessions not being able to address some of these socio-economic issues. The codes identified in the analysis for this category are presented in Table 3 below.

Table 3.

Coding frequencies for the influence of context on therapeutic effectiveness.

Code name	Number of session transcripts containing code	Number of times referenced ‘in transcripts’
Poverty	63	98
Life stressors or trauma	47	72
Physical, sexual, and economic abuse	32	37
Alcohol abuse by partners	18	21
Food insecurity	17	26
HIV/AIDS	13	16
Absent father of baby	11	18
Discrimination from community	9	10
Dropping out of school	8	14
Negative clinic experiences	8	12
Insecure housing	5	6

In a distressing example of the effects of poverty on mental health and subsequent helpfulness of counselling, one participant said she was almost driven to suicide because she was unable to buy things for the children during winter. She said, “children must get clothes, do you see, that is stress. Why would someone not want to kill herself?” (DB, S2). In another example, a participant explained that she felt symptoms of depression and anxiety because she had “no Pampers [diapers] or food for my child” (AC, S1). In a description of the level of food insecurity some participants face, another participant stated to her counsellor that she used not to have an appetite for certain foods, but now she will eat anything, “due to the lack of food availability” (BE, S1).

Physical and sexual abuse, often linked to substance abuse by partners, were frequently reported by participants before and between sessions. In addition, aspects around a lack of support from partners were also commonly described. For example, one participant explained to her counsellor that she had taken her boyfriend to court because “he wanted to sleep with me without my permission and he was beating me” (AC, S2), and another said that “the father of my baby is cheating and beating me a lot, and he uses the stones to beat me” (BD, S4). The effect of abuse on mood, despite counselling, was made clear when a participant described being abused over the weekend, in her last session, saying: “All that [being abused] has never happened, and now it is happening this weekend… I am feeling weak now” (ED, S6).

In a similar situation, another participant demonstrated economic abuse linked to a partner’s substance abuse, when she said that he ‘drank’ all the money that they had, leaving nothing for living costs. She said, “he brings money at home, but he just drinks it. So now we do not have money to even [pay for] rent. He used to bring R100 (~ US$7) home but now he does not bring anything” (BC, S3).

The presence of HIV/AIDS also affected women’s social, relational and economic participation, and was an underlying factor inhibiting recovery from depression. A few women attributed the onset of their depression to the discovery of their HIV status, as one woman said, the reason for her depression was “knowing that I am HIV positive” (AA, S1).

Instances of trauma shared by the women in the transcripts were distressingly common, and included murder, family death, and physical abuse. The following are examples of the reality that women face in Khayelitsha. One participant described the murder of her boyfriend: “His younger brother found a job and he was using alcohol and drugs. On that night my boyfriend was sleeping already, and his brother attacked him with an axe” (AB, S1). Another participant had experienced the death of three family members in her life: her mother at age 12, her older sister, and most recently her brother, who she used to stay with after her mother died, until he “shot himself to death” (BB, S1). This participant’s multiple traumas had a direct impact on her schooling, as she consequently dropped out before finishing school.

The above quotations paint a bleak picture of the nature and complexity of the problems that participants faced. These contextual examples may also be referred to as ‘persistent distress’, where the women experienced trauma and distress throughout and after the counselling process. The effect of the counselling may thus have been limited in extent or duration by the consistent presence of this distress.

3.2. Counselling factors affecting quality of delivery

The analysis found certain strategies or behaviours used by the counsellors that did not align with what was intended in the counselling manual. The most common of these across all of the counsellors were: misinterpretation of therapeutic modalities, inadequate explanations of therapeutic activities or of the required homework, inappropriate reading from the manual, minimising problems that participants identified, providing absolute assurances for solving these problems, providing advice, and reprimanding participants. These are depicted in Table 4 below.

Table 4.

Coding frequencies for counselling factors affecting quality of delivery.

Code name	Number of session transcripts containing code	Number of times referenced ’in transcripts’
Providing advice	54	70
Minimising problems that participants identified	52	59
Misinterpretation of therapeutic modalities	40	57
Inadequate explanations of therapeutic activities or of the required homework	24	28
Inappropriate reading from the manual	15	20
Absolute assurances for solving problems	10	14
Reprimanding participants	7	8
Trivialising participants’ solutions	6	7
Using personal experiences to explain concepts	2	2
Counsellor inflexibility with participant concerns	1	1

The most common of these strategies were used by all of the counsellors at different points across a variety of participants and sessions. To start, some counsellors seemed to misinterpret the therapeutic modalities that they were supposed to have been teaching to participants. This is demonstrated with an example of when a counsellor read an instructional sentence from the manual to a participant (not meant to be read aloud), and then went on to explain the concept to the participant in her own way:

“Psychoeducation: The counsellor teaches the mother about mental problems by explaining the cause, what to expect and what the solution is. Psychoeducation helps to reduce being looked down upon due to depression. Did you hear? Psychoeducation is depression. Do not be confused when I use them interchangeably” (Csr A, AD, S1).

In addition, the counsellors tended to provide inadequate explanations of how to do the session activities or of the required homework, and sometimes read inappropriately from the manual (as above), when they were expected instead to explain these in their own words.

There was also evidence of counsellors minimising problems that the participants raised in the sessions, thereby potentially invalidating the experiences of the mothers. An example of this is shown below where a counsellor did not acknowledge or follow up a participant’s revelation of pain and lack of recognition of her illness at her church:

“They know at church that I always sing even though I don’t know how to sing. Then they asked me, why am I not singing? And I would be thinking, they have no idea how much pain I am in”. Counsellor: “Okay so you say you like going to groups and singing. You can write your activities in your book. We will talk about them next week. You write the things you like” (Csr A, AF, S3).

Another strategy that the counsellors used was that of making assurances that after the six counselling sessions, the mothers would feel fully better and that their problems would be solved. One participant asked a counsellor whether ‘feeling sad’ was going to affect her baby. She answered:

“No, you have had counselling, so you are not going to have a problem. It can affect a baby if you are depressed when you are pregnant. If you talk about your feelings, nothing is going to happen” (Csr E, EA, S4).

Giving advice around how to solve problems or what to do about partners or boyfriends was a particularly common trend amongst counsellors. This is illustrated in the example below when a counsellor said to her participant:

“[DE], let me give you a piece of advice about men. When you get angry that he spends time or drinks with his friends, the first thing in his mind is that his girlfriend or wife is controlling him. Do you understand? The right medicine for them is [for you] to keep quiet and not ask anything, even if you want to talk” (Csr D, DE, S6).

Another counsellor demonstrated her feelings about men by saying:

“Now you think that he is cheating when he goes out. I can advise you that if he is buying food at home [for you], just keep quiet. All men are cheating so you need to look after your children” (Csr E, EA, S2).

Last, there were examples of counsellors scolding or reprimanding participants. The following example reveals visible frustration and scolding from a counsellor:

“The last session you spoke about your likes and dislikes right? Where is your book? You do not write [your exercises]. I gave an example for you and you did not write. You see? I even said choose here all the way to this side. You see?.. I gave an example and even made brackets for you… You did not write. Write! Write! At all. Are you getting old? … Do not think you have escaped because you still have the book [at home]. I want that book and this one full for Session 5. Do you understand?” (Csr F, FA, S4).

3.3. Elements identified by participants as therapeutically effective

Despite evidence of barriers to therapeutic effectiveness, the analysis found that throughout the sessions, participants were providing spontaneous accounts of improvement in mood, as well as examples of changes in communication, behaviour and cognition. These were demonstrated through statements such as:

“I feel a happiness that I am not used to” (AD, S4).

“I don’t think too much… I am now sleeping at night” (AE, S3).

“I am alright now, unlike when I was over-thinking” (EF, S6).

“Yes, I used to be angry and thinking negatively but now I can control myself” (EA, S5).

Importantly, beyond reporting this subjective improvement in symptoms, participants provided examples of the reasons or facilitators for these improvements, describing various elements that seemed to have been therapeutically effective in the counselling sessions. These attributions involved: improving knowledge through receiving advice, a normalisation of feelings of depression, the experience of confidentiality, trust and empathy, sharing problems, being heard, feeling a sense of relief, and re-establishing social connections. Table 5 below portrays these codes.

Table 5.

Coding frequencies for elements identified as therapeutically effective.

Elements identified as therapeutically effective: Code name	Number of session transcripts containing code	Number of times referenced ‘in transcripts’
Improving communication	53	62
Sense of connection	28	33
Sense of relief	19	21
Receiving advice	15	18
Confidentiality and trust	15	15
Being heard and experiencing empathy	14	15
Sharing problems	13	18
Normalisation	9	11
Commitment from counsellor	4	5
Having someone external to talk to	3	3

Many participants related that receiving advice was beneficial in improving their communication strategies. In western psychotherapy contexts, this strategy may not be endorsed, but it was perceived as helpful by participants themselves. For example, in her fifth session, one participant demonstrated the improvement in her communication with her partner, saying that there was a “better difference in my life because you advised me on how I should behave. Not to shout at the person I live with” (DF, S5). There were also many instances when participants asked outright for advice from their counsellors.

Psychoeducation and subsequent normalisation of depression seemed to go hand in hand. The sense of recovery that women described, having learnt about depression and its causes, appeared to lead directly to a subsequent feeling of relief and release of self-blame, guilt and internalisation of feelings. This was demonstrated in the following quotations:

“I know now about the depression and how to behave. I learned that depression is a healing process, but you need to speak out about your feelings” (AG, S6); “I feel like sort of forgiving myself a bit now… I felt that my depression went away after we spoke about that first session on depression… When we met and you taught me about what causes depression” (CB, S6); and, “I learnt not [to] internalise something when I have a problem” (BA, S4).

Although women live cheek by jowl in the crowded Khayelitsha township, they regularly experience social isolation, particularly when a potentially stigmatising event has occurred, such as rape, sexual abuse, having no money, an unwanted pregnancy, HIV diagnosis, or losing a partner. These instances of multiple trauma are often stigmatised, and many participants related that the counselling sessions provided a space for them to have someone to talk to and with whom to share their problems. Related to this was a subsequent sense of relief that women experienced after being able to talk about a traumatic experience. This seemed to be an important mechanism in improving affect amongst the participants. Some of the participants reported the following in this regard:

“Now I am not thinking about killing myself because I can share my problems” (EA, S2).

“Ever since I came into this programme, I am now able to talk and empty my chest. I can face my problems now. I just put it behind me” (AF, S6).

And, “I realised that when I talk about certain things I get healed. Talking to you made me feel better. I managed to see how I get hurt by some things, and what I am supposed to do to achieve certain things” (BB, S2).

Following the experience of sharing problems, a further therapeutic benefit of ‘being heard’ or experiencing empathy from counsellors was described by participants. Despite the instances of counsellors appearing to brush over participants’ problems, there were reports from participants about how it felt to share their problems and have someone really listen to them. One participant said:

“I don’t think the same way I used to before… Because there is someone I talk to who listens to me. Someone who doesn’t judge me. Because I also don’t know why I am that way… And not just listen, but try to help me solve the problem” (DE, S4).

The aspect of confidentiality and subsequent trust between participant and counsellor was also mentioned by participants as an important therapeutic element. The therapeutic effects were enhanced through an acceptance of the disclosures from participants, and a subsequent normalisation of behaviour and feelings.

One of the more subtle findings from the data was that participants were experiencing an increased sense of connection between themselves and their counsellors, a sense of hope, and consequently, an increased ability to improve and re-establish their own social support in their lives. The sense of connection was related by the following statements from participants to their counsellors:

“When I talk to my boyfriend, he does not want to be involved. So, at least by meeting you, through talking to you, it is better…Because sharing is important” (CB, S3); [The sessions] removed my stress. Through listening, I discovered that this way helps [better] than staying by myself and not having someone to solve problems with” (CB, S6).

“I did not think that I would be like this, [being] able to talk about a problem. I am like the only child. I talk, but I cannot talk too much… If I don’t want to I don’t want to and you won’t force me. I learnt a lot of things here and I am grateful” (AF, S6).

The subsequent effect this had on improving social support and social connection is illustrated below:

“Now I want to do good things; I hated my family members… But I want them now!” (AD, S4).

“Now I can socialise with people even though I had no relationship with them before… [I share with] my friend next door. We just share our things” (AC, S3).

“It was nice to see the neighbours, rather than being alone and being locked up… To be out with friends was nice” (BE, S3).

4. Discussion

A grounded theory analysis of the AFFIRM-SA counselling session transcripts provided a valuable method for evaluating the intervention and created an opportunity to ‘see in’ to the sessions, rather than eliciting responses from participants or counsellors after completion of the intervention. This approach revealed many important therapeutic and contextual aspects of the intervention that were not identified through the outcome assessments for the RCT and offers important learnings for future research and intervention development.

The findings present apparently contradictory narratives in which there were on one hand, factors that compromised the quality and efficacy of the counselling, and on the other hand, elements of the intervention that were identified as therapeutically effective by the study participants. These nuances reflect broader findings in global mental health research, whereby task sharing works for certain populations, some of the time; yet there are important exceptions, from which lessons can be drawn on narrowing the treatment gap for perinatal depression in resource constrained environments (Padmanathan & De Silva, 2013).

4.1. Influence of socio-economic context

This study found that the socio-economic context of the intervention influenced how the intervention was received by participants (Kok et al., 2015). The context in Khayelitsha is marked by poor physical conditions (such as inadequate health services, sanitation, roads, electricity, housing and high cost of nutritious food), and highly stressed social conditions (such as unemployment, absent family members, partners having multiple girlfriends, relationship insecurity, and financial insecurity). These factors influenced the extent to which the counselling sessions were able to improve women’s mood, as the participants were not able to separate themselves from the circumstances that were, to a large extent, causing their depression in the first place. A review of explanatory models of depression in Sub-Saharan Africa found that participants identified depression symptoms as only one problem among many, and that these were consistently ranked below the economic problems that they faced. They found that depression was attributed largely to social adversity, economic and relationship problems, and HIV-related issues (Mayston et al., 2020).

Burgess et al. (Burgess et al., 2020) acknowledged this in a commentary articulating the difficulty for psychosocial interventions to improve mental health and wellbeing against the backdrop of ‘intractable’ social, economic and political determinants of ill health, many of which are an every-day reality of the lives of people in LMICs living in poverty. Similar to AFFIRM-SA, past psychosocial interventions have not been able to adequately deal with these broad determinants or respond to participants’ circumstantial needs (Padmanathan & De Silva, 2013). Interventions may indeed require a very specific interaction of activities, setting, and participant histories to produce an outcome intended by those designing the intervention (Murdoch, 2016). It is therefore a possibility that despite the various factors identified as helpful by the participants, the nature of the context and the mechanisms of poverty were often too powerful for therapeutic strategies to have any lasting effect.

4.2. Quality of delivery: counselling factors

Second, this study found that various counselling factors or tendencies that counsellors used in the intervention meant that the counselling was not always delivered as intended in the original protocol.

Various factors demonstrated this, such as a misinterpretation of therapeutic modalities, inadequate explanations of therapeutic activities or of the required homework, inappropriate reading from the manual, minimising problems that participants identified, providing absolute assurances for solving these problems, providing advice, and reprimanding participants.

Some reasons for this may have been that the counsellors did not achieve optimal proficiency in the techniques required of them, and that there were too many modalities in the intervention to master and subsequently teach to participants themselves. Other South African studies that have examined competency in mastering therapeutic techniques, such as Motivational Interviewing (MI), also found that CHW-counsellors did not meet the target competencies in the techniques (Dewing et al., 2013a; Evangeli et al., 2009). These findings highlight the importance of competency, training and supervision of CHWs in the provision of psychosocial interventions. Indeed, many of the above counselling factors may have been remediated by initial training longer than the five days that were provided, and the inclusion of refresher training sessions throughout the trial. Dewing et al. (Dewing et al., 2013a) also questioned the ability of CHWs to deliver an MI intervention following only five days of training. In addition, despite weekly supervision provided in the trial, it may have been enhanced by the addition of role plays and more frequent assessments of competency in the various therapeutic modalities. The ‘Ensuring Quality in Psychological Support’ (EQUIP) initiative has been proposed as an approach toward competency-based training and supervision (Kohrt et al., 2020). This online platform aims to aid the selection of CHWs and identify certain skill targets for them, which would then inform further training design and content.

Another factor that may have played a role in the occurrence of counselling tendencies affecting the quality of delivery is that the counsellors were not removed from their own cultural and community norms and expectations of what constituted psychological “counselling”. Notable in this study was the presence of advice-giving by counsellors to participants, particularly relating to their own belief systems around relationships, gender roles, pregnancy, and community attitudes and stigma. Many participants also actively asked their counsellors for advice. It is therefore important to consider that some of the strategies that the counsellors used in this context might in fact have been appropriate, and consistent with social norms in this context. The provision of contextual advice by CHWs, and the positive role this played, was also reported in studies in Zimbabwe, Nepal, Burundi and South Africa (Chibanda et al., 2017; Dewing et al., 2013b; Jordans et al., 2012; Kohrt et al., 2015).

Although the Western client-centred model encourages counsellors not to give any advice and rather elicit responses from participants, this element of counselling should be explored and potentially be incorporated into context-specific interventions rather than negated and discouraged. In order to properly understand and support CHWs in South Africa, we need to conduct a broader analysis of the health system and political context (Colvin & Swartz, 2015). Murdoch (Murdoch, 2016) writes of the ‘historical context’ that trials are situated within, in the recognition that “moments of intervention delivery are inextricably linked to wider social forces” (p. 11) that will affect how the intervention is delivered. A deep understanding of community structure, history and services, is thus integral to implementing psychological services through CHWs in their communities (Thompson et al., 2012).

4.3. Therapeutic elements

Last, certain factors were identified by participants in their sessions as being helpful in improving mood. These included receiving advice, normalisation, sharing problems, a sense of relief, learning communication skills, the experience of confidentiality and trust, of empathy, being heard, and establishing connection and social support. What is interesting is that all of these factors corroborate elements already identified in the growing body of research on “common elements of therapy” (Cuijpers et al., 2019; Singla et al., 2017). These elements are universal across experiences and modalities of therapy, and are vital in determining the outcomes of treatment (Barth et al., 2012). They include communication skills, empathy, collaboration, emotional regulation, therapist-client alliance, and involvement of a significant other (Kohrt et al., 2020; Singla et al., 2017). Although common elements were not formally included in the AFFIRM-SA intervention, the nature of the unfolding therapeutic relationship appears to have enabled some of these to emerge.

The findings endorse what various authors are now proposing: that the use of a limited number of common elements may assist in addressing the large treatment gaps in LMICs, through incorporating them as key therapeutic factors in the implementation of mental health interventions (Kohrt et al., 2015; Murray & Jordans, 2016; Singla et al., 2017). This was recently demonstrated specifically for perinatal depression in India and Pakistan through a mediation analysis where authors found that perceived social support and patient behavioural activation mediated the effects of the ‘Thinking Healthy Programme’ across diverse contexts (Singla et al., 2021).

5. Recommendations

This section uses the findings from the process evaluation of the AFFIRM-SA RCT to formulate recommendations for the design and implementation of research and practice in task-shared psychosocial interventions in LMIC.

5.1. Conduct participatory formative research

To begin with, there is value in going beyond formative work traditionally conducted for RCTs (such as qualitative interviews) and conducting more participatory formative research, as a means to inform and improve the development of counselling interventions (Bolton, 2019; Burgess et al., 2020; Petersen et al., 2012). This process can assist in gaining insights into the roles of the health system and the individuals involved, and give a voice to counsellors and those with mental illness (Frymus et al., 2013, pp. 1–5). Employing a deep participatory process in the initial development of an intervention with the counsellors themselves is important to create a sense of ownership of an intervention and empowerment of counsellors, which in turn may encourage an acceptance and understanding of a counselling intervention in a community and increase sustainability in the longer term (Chibanda et al., 2017).

5.2. Respond to contextual need

It is essential that interventions take into consideration the broader context within which they are implemented (Kok et al., 2015), such as social, cultural and gender norms, values and belief systems, education and health provision systems, and political and economic context and history. This includes understanding local conceptualisations of depression and means of expressing symptoms of mental illness, as explored in the formative research for the AFFIRM-SA trial (Davies et al., 2016). In the intervention design stage, this would also include incorporating context-specific strategies suggested by CHWs, and indigenous solutions to problems (Rahman, 2007).

It is also important to respond to the contextual needs expressed by participants themselves (Burgess et al., 2020) such as for economic skills, improved food security, and safety from violence and abuse. This endorses the idea of including economic skill building components and strategies to deal with intimate partner violence (IPV) in task-shared interventions for mental health, to realistically address the context that women in low-income settings face. Nakimuli-Mpungu et al. (Nakimuli-Mpungu et al., 2020) provide an example of a psychotherapeutic intervention in Uganda where participants were consulted on what they felt would be culturally appropriate means of counselling, and what they needed from counselling. The subsequent group intervention included basic CBT techniques, identifying social support, and income-generating skills, and demonstrated significant improvements on depression symptoms.

5.3. Training and supervision

The current study also highlighted the necessity of in-depth training and supervision. The AFFIRM-SA RCT provided five days of training for the counsellors, but in retrospect this was not long enough to teach them the various therapeutic modalities and techniques as well as how to use the manual. A key recommendation is that training for counsellors is conducted over at least three weeks, and that refresher training sessions are conducted throughout the intervention. Furthermore, co-production of the intervention with the counsellors would reduce training needs. Barnett et al. (2018) write that training outcomes improve when there are opportunities to practice and receive feedback on skill use which longer training and refresher sessions would allow for.

It is also important to examine therapeutic competency of CHWs continuously when conducting psychological counselling. The ‘EQUIP’ initiative may assist in facilitating competency assessments and evaluate therapeutic training for CHWs (Kohrt et al., 2020). In addition, CHWs should ideally hold some basic interest or inclination in mental health and counselling and have a desire to help those with mental illnesses (Spedding, 2017).

Supervision is similarly vital for effective implementation of interventions (Rahman, 2007), counsellor adherence, skill development (Beidas et al., 2012), monitoring of fidelity, provision of a support structure, and as a predictor of behavioural change (Murray et al., 2011). Supervision in this trial took the form of group discussions of case studies, participant progress, and referrals. Adding regular assessment and discussion of the counsellors’ understandings of the therapeutic modalities and role plays to demonstrate this, may have improved therapeutic effectiveness of the RCT. Other recommended monitoring strategies for low-resource settings include counsellor self-report, live observation of sessions, role-play behavioural rehearsal (Murray et al., 2011), and peer-supervision using a structured scale (Singla et al., 2020). The findings also highlight the importance of training in supervision skills for supervisors (Scott et al., 2000). Adequate supervision is equally critical to monitor and relieve emotional burnout and secondary trauma that the CHWs may experience in their counselling (Barnett et al., 2018).

5.4. Adopt a staged or stepped care approach to treatment

There is a possibility that in the current RCT, the counselling sessions assisted women who were distressed, but not necessarily those who were clinically depressed. Future interventions may therefore benefit from a distinction between feelings of distress and clinical symptoms of depression, and to adopt a ‘staged’ approach (Patel, 2017), and providing more appropriate strategies for those in different clinical stages (Patel, 2017). These would range across a spectrum from peer support groups at local clinics on one end, to individual psychological therapy on the other.

Stepped care is another element that can be employed in resource-poor environments (Hanlon et al., 2014). This can involve a number of elements, beginning with, for example, psychoeducation for any at-risk individuals, and then group psychosocial interventions for anyone identified with moderate to severe mental disorders, with referral to specialists only for suicidality or severe and unmanageable disorders (Davies et al., 2019b). This mode can improve efficiency and capacity to treat more people in LMICs (Hanlon et al., 2014).

5.5. Include common elements of therapy

Following intensive formative research, the design of the intervention should include the use of common elements that have been identified in international literature and in local studies, if these are available (Cuijpers et al., 2019; Murray & Jordans, 2016; Singla et al., 2017). Examples from this study include establishing a sense of trust and confidentiality, creating an environment that fosters problem sharing and a willingness to talk openly to someone, increased awareness and education, normalisation of depression, and improving communication skills. Other elements previously identified include eliciting social support, emotional regulation, collaboration, empathy, active listening, involvement of a significant other, therapeutic alliance, active problem solving, and family involvement (Jordans et al., 2012; Singla et al., 2017). Researchers recommend that counsellors be trained in these common elements before they are taught specific treatment skills (Jordans et al., 2012; Kohrt et al., 2015), and that they should form the basis for all psychological treatments, regardless of treatment modality (Singla et al., 2017).

5.6. Use a single therapeutic modality

Given the educational levels of both counsellors and participants, using only one therapeutic modality and repeating it over a number of sessions may decrease training requirements, increase counsellors’ competence and confidence, and encourage a deeper understanding of a particular skill set for participants. This may also ease the difficulty of counsellors having to split attention between reading a manual and establishing therapeutic alliance with participants. Examples of successful interventions using single modalities include the CBT-based ‘Thinking Healthy’ trial in Pakistan (Rahman et al., 2008) and the problem solving ‘Friendship Bench’ trial in Zimbabwe (Chibanda et al., 2016). In addition, even when using evidence-based approaches, it is still vital to adapt them to local needs (Bolton, 2019).

5.7. Use iterative research and implementation science

In AFFIRM-SA, the limitations of working within a strictly controlled RCT design hindered opportunities for counsellors to give feedback on the design, content, or contextual need of participants once the intervention was underway. One means of improving this would be to use a more iterative and responsive approach to intervention design and implementation. This might be achieved through a ‘continuous process evaluation approach’ (Murdoch, 2016; Padmanathan & De Silva, 2013), whereby interventions are qualitatively assessed before, during and after the intervention, and feedback is obtained from counsellors and supervisors on both the content and design of the intervention. Importantly, these evaluations need to be budgeted for and included from the beginning stages of research proposals and grant writing, with funders building in expectations of this process.

Murdoch’s continuous process evaluation framework (Murdoch, 2016) offers a strategy for operationalising the evaluation of implementation fidelity and theoretical fidelity. This would be achieved by: 1) Setting out macro, meso and micro contextual features before trial implementation (possible inclusion of these in a theory of change model); 2) Targeting where tensions are likely to occur between different contextual features, including a consideration of the trial protocol as a contextual feature itself; 3) Searching for ‘disruptions’ in targeted activities (e.g. where counsellors are not following protocol); and 4) Considering the consequences of these disruptions for how the trial is conducted and the implications of these consequences (Murdoch, 2016).

The use of implementation science research also provides an effective means of examining implementation factors (Curran et al., 2012). This methodology involves a responsive approach throughout the intervention. Singla and Hollon (Singla & Hollon, 2020) set out three important criteria for implementation research and the examination of how treatments can be effective. These are: 1) an ongoing examination of relevant implementation factors related to the therapist and treatment; 2) moderation, mediation and cost-effectiveness analyses to assess strengths and challenges of interventions; and 3) linking therapist competence with treatment outcomes, using digital methods to ensure acceptable training and supervision. These may complement traditional treatment effectiveness analyses. Conducting these examinations is necessary to help scale up psychological treatments and subsequently, to better address the burden of CMDs globally.

5.8. Reporting

Last, it is vital that all stages and processes of task-shared research be reported, including the training, supervision, and intervention development processes (Damschroder et al., 2009). Feasibility studies should be reported alongside trial outcomes, so that feasibility and acceptability is more fully realised than only through a pilot study (Hoddinott, 2015). For randomised controlled trials and their pilots, the Consolidated Standards of Reporting Trials (CONSORT) guidelines should be used (Schulz et al., 2010).

6. Limitations

There are a few limitations to this study. First, the data for the grounded theory analysis was retrieved from counselling session transcripts rather than in-depth interviews which ask more directed or specific questions. This allows for a possibility of over- or misinterpretation of the data. Nevertheless, the use of this method allowed the coding and categories to emerge from the data itself rather than be imposed on it, which to a certain extent should have negated this possibility. Second, due to the nature of grounded theory, the data were rated and analysed only by one rater, and inter-rater reliability of the identified codes could not be assessed. This limitation was compensated through a workshop with an expert panel to provide feedback on the reliability and acceptability of the codes. Third, an analysis of counselling transcripts from participants who did not complete all six sessions may have identified possible reasons for participants not completing sessions that related to counselling style and participant response. In addition, these participants may have reported fewer benefits or positive outcomes from the sessions, and this could have been a reason for non-completion. Contrastingly, the 39 participants in this study may have completed all six counselling sessions because they were obtaining some benefit from them, leading to inflated responses of the positive outcomes of the sessions. Last, an indepth analysis of the styles particular to each counsellor and issues of adherence from their participants was not conducted as it would have required a more specific focus on each counsellor and was beyond the scope of this paper.

7. Conclusion

This study explored the nuances of implementing task-shared psychosocial counselling for perinatal depression in Khayelitsha, and discussed the ensuing recommendations for implementing future interventions in other low-income settings. A grounded theory analysis of the counselling transcripts allowed for an in-depth understanding of how counsellors implemented the sessions, how participants experienced the sessions, what outcomes they reported, and what factors influenced delivery and reception of the counselling. To our knowledge, this type of evaluation of a task-shared intervention has not been conducted before. The narrating of these processes also aligns with reporting recommendations for interventions articulated by authors such as Singla and Hollon (2020), Moore and Evans (2017), and Damschroder et al. (2009).

Critically, the findings emphasise the importance of adequate training, supervision, and monitoring of quality, in interventions using CHWs to provide psychological therapy (Barnett et al., 2018; Verhey et al., 2020). They also highlight the influential role of poverty-related risk factors in experiencing depression and hindering therapeutic effectiveness. This confirms that these broader determinants of mental illness need to be addressed if symptoms are to be improved (Burgess et al., 2020). This could involve designing multi-component interventions that address mental illness at multiple levels, and implementing holistic social assistance programmes and approaches that attend to a range of poverty related factors such as unemployment, deprivation, disempowerment and food insecurity. Formative participatory research for intervention development (Burgess et al., 2020), and implementation of a staged approach to measuring and addressing depression (Patel, 2017) would help to better incorporate the voices and needs of all stakeholders.

This study corroborated other research demonstrating the effectiveness of common elements in therapeutic settings (Singla et al., 2017), such as the experience of empathy, trust, being heard, and feeling supported. This adds to the growing evidence proposing the use of these elements in training and implementing task-shared therapy, regardless of the type of modality used. In this context, participants additionally gave value to receiving helpful advice from their counsellors and experiencing a sense of connection with them.

There are numerous challenges in implementing task-shared mental health interventions, particularly in the directive that they should be cost effective, while also increasing coverage of services at a primary care level. Process evaluations are invaluable in going beyond the assessment of whether an intervention works, into reasons for why it works or does not work (Cuijpers, 2019). This study presents a number of contextual considerations and therapeutic elements relevant to designing and implementing more acceptable and responsive global mental health interventions in LMICs.