Background
Perinatal depression is a significant public health issue in both high income (HIC) and low and middle income countries (LAMICs) [
1,
2]. Perinatal depression which refers to the experience of depression during pregnancy and up to one year post-partum can be associated with adverse consequences for both mother and baby [
3]. In LAMICs, where resources are few, and access to mental health professionals is limited, [
4‐
7] prevalence of perinatal depression is estimated to be 15.9% [
8]. The experience of perinatal depression in LAMICS is exacerbated by poverty, unemployment, HIV/AIDS, and intimate partner violence.
LAMICS have a “treatment gap” where up to 75% of people who need mental health treatment do not always receive optimal care [
9,
10]. Research suggests that task sharing is a successful means of addressing this “treatment gap” for perinatal depression in resource poor settings [
11,
12]. Task sharing is an approach to mental health service provision whereby non-specialist health workers provide care for less complex cases under the training and supervision of a specialist. This shares the burden of care [
4] while providing locally relevant interventions to people from the same community and cultural background who speak the same language [
13,
14]. Although the line between efficacy (whether the treatment works under ideal circumstances) and effectiveness (whether the treatment works in real world situations) [
15] in relation to task sharing for mental health in LAMICS has been somewhat blurred, many trials have proceeded to effectiveness evaluations without necessarily demonstrating efficacy. Several studies have suggested that task sharing has benefits and results from a Cochrane review indicate that non-specialist workers can be trained to deliver psychological interventions with training and supervision in order to improve the symptoms of perinatal depression in mothers [
16]. Task sharing has also been shown to be effective for the treatment of perinatal depression in Pakistan and for depression in men and women in Uganda [
17,
18]. There is a need for both process evaluations and in-depth qualitative analyses of task shared interventions to develop a better understanding of factors contributing to their sustainability. Providing qualitative evidence on interventions is crucial for gaining insight into the participants’ and service providers’ views on the development of acceptable interventions [
19].
The new United Kingdom (UK) Medical Research Council (MRC) framework for process evaluations provides guidelines for conducting process evaluations in order to assess the quality of implementation and fidelity to the intervention [
20]. The framework further recommends examining the relationship between three main factors: implementation, context and mechanisms [
20].
Implementation includes examining the resources provided for the intervention and their appropriateness, such as counsellor training, supervision, manuals, dose and reach (the total number of sessions and participants reached) [
20]. The
context includes examining the external environmental or community (such as rural or urban setting and common cultural or religious practices), and service structure factors such as acceptance by local Primary Health Centre [
5].
Mechanisms refer to participant responses to the intervention and the aspects of the intervention that lead to change in the participant’s behaviour including counsellor motivation to conduct the intervention and participants’ motivation to attend sessions [
20]. In this study we understand mechanisms to include both the mechanisms of the intervention and the mechanisms of implementation (which are important to consider in the context of task sharing). The context, implementation and mechanisms can be used to examine factors that affect the intervention [
20]. There have been previous qualitative studies on task sharing for mental health care in LAMICS without a clear process evaluation which highlight important factors that affect the intervention’s acceptability and feasibility. These include: service providers’ level of confidence, distress experienced by participants, fidelity to the intervention, acceptability of the intervention, costs and policy alignment and adequate incentives [
21]. Several barriers to task shared interventions have been noted including poor adherence, low acceptability of talk therapy, stigma of mental health interventions and burnout due to increased workload for service providers [
13]. Synthesising themes across these various studies is useful to evaluate the appropriateness, acceptability and effectiveness of interventions [
22,
23]. Qualitative studies can provide nuanced detailed understandings regarding the process of delivering interventions which are not accessible through quantitative data. Within the context of task sharing, qualitative studies can complement quantitative studies because most studies do not report qualitative data from trials, and this is an important area to highlight for future research [
24].
To our knowledge no systematic review has been conducted to synthesise qualitative evidence on process evaluations of task shared intervention for perinatal depression in LAMICs. This review seeks to answer two main questions: (i) to what extent are qualitative process evaluations conducted on task shared interventions for perinatal depression in LAMICs; and (ii) what is the best way to synthesize emergent themes from the process evaluations with the MRC framework for conducting process evaluations [
20]?
Methods
Search strategy
Five electronic databases were searched between September and December 2015 - Medline/ PubMed, PsycINFO, Scopus, Cochrane Library and Web of science. The search terms included four concepts (a) “maternal depression’” (b) “intervention” (c) “lay counsellor” which were expanded by using “community health worker” OR “non-specialist” and (d) “LAMICs” as determined by the World Bank Country classification. These phrases were adapted for use in each database.The terms “task sharing” and “process evaluation” were excluded from searches since they restricted the number of abstracts identified. In PubMed the following search terms were used, and adapted for use in other databases:
(((((perinatal) OR prenatal) OR antenatal) OR postnatal) OR postpartum) OR post-partum AND depression AND (((((community health workers) OR community health aides) OR village health workers) OR health personnel) OR fieldworkers AND counselling OR psycho-social intervention*AND developing countries.
A full description of the search strategy is included as Additional file
1. The inclusion and exclusion criteria are presented in Table
1.
Table 1
Inclusion and Exclusion Criteria
Publication Type | • Qualitative evidence of process evaluations of psycho-social treatment interventions for antenatal or postnatal depression | • Quantitative studies which do not have a qualitative component |
Study Design | • Studies which evaluate effectiveness of both pharmacological and psycho-social intervention | • Studies that only evaluate pharmacological interventions |
Condition of Interest | • Antenatal OR Post-natal OR Perinatal depression | • Studies of other conditions which are not perinatal depression |
Type of intervention | Psycho-social counselling or psychoeducation | • Studies that do not include counselling or psychoeducation |
Time point | • Post-intervention evaluation | • Pre-intervention evaluation |
Study Population | • Group and individual intervention by non-specialists | • Studies where intervention is conducted by mental health specialists |
Intervention Location | • Studies in LAMICS | • Studies in HICs |
Language | Studies in English | Studies not in English |
Abstracts identified were imported into Endnote and duplicates were removed. MM (primary reviewer) and SM (secondary reviewer) independently reviewed the abstracts for each paper using the eligibility criteria described in Table
1. Upon initial screening, the majority of the articles were excluded for the following reasons; not an intervention study, a review paper not intervention and treatment other than counselling. Once full-text articles had been retrieved, MM and SM independently reviewed the studies again and the following criteria were used to further exclude papers such as: studies that do not employ a qualitative methodology, were not process evaluations nor task shared interventions and did not target perinatal depression. MM and SM had several face to face discussions to reach consensus on studies. In cases where studies provided limited information on the intervention, the authors were contacted to provide further information.
Quality appraisal
The review used the 2009 Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement which ensures that the study reported fits the reporting standards of systematic reviews, assesses the quality, structure and whether there is a clear explanation of the objectives, methods and results [
25]. The PRISMA Statement is included as Additional file
2. Data were extracted using a standard data extraction table which included the following: date of publication, setting of the study (hospital/ clinic/community), study design, number of participants, age range, measures used, validity of measures, quality assessment and main process evaluation findings. This table is provided in Additional file
3. The quality of the included studies was assessed by both reviewers independently using the Critical Appraisal Skills Programme (CASP) checklist which examines risk of bias, and whether the study design, recruitment strategy, data collection and analysis were appropriate for the study [
26]. The CASP checklist is provided in Additional file
4.
Data analysis was conducted using thematic analysis. The reviewers followed the 3 steps set out by Thomas and Harden [
27]: (i) free coding of data (ii) organising coded data into descriptive themes and lastly, (iii) generating analytical themes. The reviewers read the full text articles and conducted free coding of data by reading each line of text and organising the free codes into hierarchical groups of descriptive themes based on their similarities or differences [
27]. Meta-synthesis involved interpreting, integrating and inferring the process evaluation elements from all the included studies identified and generating hypotheses based on these findings after discussion and consensus among the reviewers [
23]. Emerging themes were integrated into the MRC Framework of context, implementation and mechanisms and further classified into sub-themes where applicable.
Discussion
This review highlighted evidence from the qualitative process evaluations using the MRC framework to examine the context, implementation and mechanisms of the interventions from three studies. The few articles included in this review highlight the paucity of evidence on qualitative data from process evaluations on task sharing interventions for perinatal depression in LAMICs.
The
context of implementation highlighted cultural aspects of the participants for all three studies in terms of access to the intervention and intervention delivery [
20]. The rural communities in Pakistan [
28] (THP) and India [
29] (Ekjut) used communal methods of intervention delivery such as inclusion of family members and other community members. The same two studies also made use of stories or illustrations to include illiterate participants and to reduce the stigma associated with depression. Both the THP and Ekjut studies emphasise the importance of observing important cultural practices in order to provide culturally sensitive interventions which concurs with Chowdhary and colleagues who suggest that cultural sensitivity improves the acceptability of interventions [
13]. It is important to note that the IPT study only collected their evaluation data from interviews with clients whereas the THP and Ekjut studies also included interviews with people who did not participate in the intervention [
28,
29].
When looking at
implementation the three studies highlight common evidence based task sharing interventions in mental health which are CBT, IPT and psycho-education. The training and supervision of the interventions varied, depending on the contextual factors. There is little information provided about the supervision of facilitators for the Ekjut and IPT studies but the THP study gives details of intensive supervision process which included discussion of problems and brainstorming solutions. The duration of the interventions varied across the studies with the IPT study providing only three sessions while the THP and Ekjut studies delivered 16 and 20 sessions respectively. None of the three studies indicate how implementation fidelity was monitored, examining the fidelity to the intervention can help researcher to see if the intervention is implemented in the way it was intended [
37].
Regarding the mechanisms of the intervention all three studies reported positive feedback about the intervention from the recipients of the intervention. Several factors appeared to contribute to the perceived effectiveness of the interventions. Intervention related factors such as the content and understandability, counsellor factors such as facilitating trust and motivation to conduct the intervention and participant factors such as motivation to attend the sessions and willingness to learn and change their behaviour, in terms of how they look after their children and relate to other people.
The factor of trust was emphasised in the Ekjut and THP studies. Trust was fostered through aspects such as combining participants and lay health workers from the same community and using cultural inclusion. Most of the THP health workers felt that the programme gave their work structure, made them more effective and that it was not a burden to their work [
28]. The Ekjut facilitators reported feeling motivated to help change behaviour of participants and also felt that trust had been developed when participants started practising what they had learnt from the groups [
29]. These findings show that motivation to deliver or attend an intervention can be seen as provider and participant mechanisms. We can see that participants view intervention positively when personnel delivering the intervention speak the same language, and that the intervention is educational and uses some form of imagery consistent with local cultural meanings [
13]. Process evaluations are helpful because they can help to increase the acceptability of an intervention. For example, the THP study made several changes based on the feedback from their qualitative interviews from their formative work such as ensuring that the terminology was appropriate.
Overall, the systematic review highlights qualitative evidence on task shared interventions which can be linked to the MRC framework categories of context, implementation and mechanisms. Understanding how the three factors relate to intervention delivery is the key to developing future interventions which are culturally appropriate and feasible in LAMICS. The context of the interventions determines the type of personnel and activities that are deemed appropriate as seen in all three studies. Counsellor factors such as motivation to deliver the intervention and facilitating trust help to encourage intervention recipients and intervention factors such as the use of visual aids and understandability of the content facilitate learning in participants and help meet their learning needs and expectations. All these factors make interventions culturally appropriate [
13].
Recommendations
For policy makers, we recommend the use of task sharing psychosocial interventions that are culturally adapted through paying attention to the needs of providers and recipients alike. It is also important to pay attention to the duration of training and mechanisms such as trust which is built over time. Therefore it is important to invest sufficient time in training, supervision and delivery of interventions.
For researchers, it is important to publish more comprehensive qualitative process evaluations following the MRC guidelines in order to aid the development of future interventions. There is limited information specifically focusing on training, supervision and monitoring of fidelity of interventions from the selected studies. This information would be helpful for the replication of the study in other LAMICS. Gaining an in-depth understanding of participant and provider perspectives is useful for the development and evaluation of interventions and applying the MRC framework in process evaluations could yield more effective results.
For lay counsellors we recommend that they be open to discussing the challenges or facilitators that they experienced when delivering interventions as this information is crucial for implementation research. For depressed women in the communities we recommend that additional support and training as peer educators be conducted in line with the recent peer-delivered THP study in Pakistan and India [
39].
Limitations
It is important to be aware of the possibility of publication bias for all identified studies, since we did not include unpublished studies and studies which were not in English. This aspect could limit the potential number of studies included in the review. It would have been helpful to know which aspects of the intervention LHW and participants found to be useful to help us understand the mechanisms involved in the effectiveness of the interventions. We contacted authors of the THP and IPT studies requesting more information on fidelity to the intervention, and training and supervision of personnel. The authors responded however the information that they provided did not shed any new light on these areas as this information was not included in their analyses. We also checked the reference lists of included studies for additional sources of information however no additional information was obtained.
Conclusion
This review highlights qualitative evidence of process evaluations for task shared interventions for perinatal depression in LAMICS from three studies. There are common mechanisms which can be recommended for successful implementation of interventions, including counsellor factors, intervention factors, and participant factors. More qualitative and comprehensive process evaluations of task shared interventions for perinatal depression are necessary to help us to understand what works and what does not work when implementing a task shared intervention both at the level of the client-provider interaction and the services and systems level. A more comprehensive application of the MRC framework for process evaluations of complex interventions would provide further information, such as fidelity to the implementation of the intervention.
Acknowledgements
We would like to thank the following collaborators who contributed to the research article:
Erica Breuer and Carrie Book- Sumner. We would also like to thank the Faculty of health sciences Evidence-Based Medicine Research Support Unit (EBMRSU) for support through the systematic review lecture series and Mary Shelton the health sciences librarian for her help with refining the search terms.