Protocol for process evaluation of ARTEMIS cluster randomised controlled trial: an intervention for management of depression and suicide among adolescents living in slums in India

Introduction There are around 250 million adolescents (10–19 years) in India. The prevalence of mental health-related morbidity among adolescents in India is approximately 7.3%. Vulnerable subpopulations among adolescents such as those living in slum communities are particularly at risk due to poor living conditions, financial difficulty and limited access to support services. Adolescents’ Resilience and Treatment nEeds for Mental Health in Indian Slums (ARTEMIS) is a cluster randomised controlled trial of an intervention that intends to improve the mental health of adolescents living in slum communities in India. The aim of this paper is to describe the process evaluation protocol for ARTEMIS trial. The process evaluation will help to explain the intervention outcomes and understand how and why the intervention worked or did not work. It will identify contextual factors, intervention barriers and facilitators and the adaptations required for optimising implementation. Methods Case study method will be used and the data will include a mix of quantitative metrics and qualitative data. The UK Medical Research Council’s guidance on evaluating complex interventions, the Reach, Efficacy, Adoption, Implementation and Maintenance Framework and the Affordability, Practicability, Effectiveness and cost-effectiveness, Acceptability, Safety/Side Effects and, Equity criteria will be used to develop a conceptual framework and a priori codes for qualitative data analysis. Quantitative data will be analysed using descriptive statistics. Implementation fidelity will also be measured. Discussion The process evaluation will provide an understanding of outcomes and causal mechanisms that influenced any change in trial outcomes. Ethics and dissemination Ethics Committee of the George Institute for Global Health India (project number 17/2020) and the Research Governance and Integrity Team, Imperial College, London (ICREC reference number: 22IC7718) have provided ethics approval. The Health Ministry’s Screening Committee has approved to the study (ID 2020-9770). Trial registration number CTRI/2022/02/040307.


INTRODUCTION
India has about 250 million adolescents aged 10-19 years, comprising nearly one-fifth of India's population. 1 The burden of mental health problems in adolescents is a growing concern globally. 2In India, the prevalence of psychiatric disorders among adolescents is estimated to be about 7.3%. 3Self-harm and depressive disorders are the leading cause of death and disability in the age group 15-19 years. 4ental health interventions for adolescents are important as 50% of adult mental disorders have an onset before the age of 14 years. 5Marginalised populations, such as adolescents living in slum communities, are particularly vulnerable because of additional

STRENGTHS AND LIMITATIONS OF THIS STUDY
⇒ This study will use existing implementation science theories and frameworks and qualitative and quantitative data will be triangulated to arrive at a comprehensive understanding of the intervention.⇒ This study will be guided by the Medical Research Council framework for developing and evaluating complex interventions and the case study method will also be used.⇒ Cases will be purposively selected based on maximum variation approach.Both intervention and control sites will be selected as cases to the enable comparison, to understand contextual factors and to avoid the Hawthorne effect.
Open access stressors related to poor living conditions, financial stress and poor access to support services. 6 7here has been limited research to test communitybased mental health interventions for adolescents living in slums in India.A scoping review of mental health interventions among adolescents in India found that of the 11 interventions included in the review, 9 were school based, 1 community based and 1 was digital. 8ost of the school-based programmes used a life skills curriculum that resulted in improvements in depressive symptoms and overall mental well-being.The review recommended the need for more interventions for early and out-of-school adolescents to ensure that the most vulnerable adolescents were not missed out.An intervention to build mental health and resilience delivered by community-based peers among highly disadvantaged young women living in urban slums in Dehradun, a city in north India, showed sustained improvements in anxiety and depression and attitudes to gender equality among study participants. 9A randomised controlled trial of a 5-month resilience-based programme among rural adolescent girls through government schools in the state of Bihar, India, delivered by local women showed that girls receiving the intervention (vs controls) had better emotional resilience, self-efficacy, social-emotional assets, psychological well-being and social well-being. 10nother school-based pilot study of of POD (Problems, Options, Do it), titled Adventures (a smartphone-based blended problem-solving game-based intervention for adolescents with or at risk of anxiety, depression and conduct difficulties) was helpful in managing their problems and stress and improving the mental health of 13-19 years enrolled in secondary schools in the Indian state of Goa. 11The Adolescents' Resilience and Treatment nEeds for Mental Health in Indian Slums (ARTEMIS) cluster randomised control trial (cRCT) is testing a community-based intervention to improve the mental health outcomes for adolescents living in urban slum clusters in India.
This paper presents the protocol for process evaluation of the ARTEMIS cRCT.ARTEMIS is a communitybased cluster randomised control trial that aims to reduce depression and the risk of suicide among adolescents living in slums.The intervention will use a mental health stigma reduction campaign with adolescents and parents of the study cohort to improve attitudes towards mental health and improve help seeking.A technologyenabled strategy will be employed for screening, clinical diagnosis and management of mental health problems (depression, other significant emotional or medically unexplained complaints and suicide risk) among adolescents by primary care doctors and community-based nonphysician health workers (NPHWs) (described in detail below).
Process evaluations provide critical inputs in understanding how interventions work in particular contexts and thus, support implementation planning beyond the trial setting. 12

Aims
The aims of the process evaluation are to: 1. Assess implementation fidelity and understand how the intervention was implemented.2. Identify key contextual factors that impact intervention delivery and outcomes.3. Understand perceptions of different stakeholders about effectiveness, acceptability and scalability of intervention components.4. Identify key facilitators and barriers in implementation of the intervention.5. Explain any adaptations to the intervention or intervention refinement during the study and their possible impact on the outcomes.

Conceptual framework
The process evaluation will be guided by the Medical Research Council (MRC) framework for developing and evaluating complex interventions. 13 14The framework highlights the importance of examining the implementation, the mechanism of impact and their interaction with contextual factors to better understand how and why an intervention does or does not work.A recent update to the framework recognises that complex interventions have several phases including intervention development, feasibility assessment, implementation and evaluation, which may not always be sequential. 15It recommends six core areas of inquiry at each phase before moving on to the next phase.They include (1) the intervention and its interaction with context; (2) the programme theory; (3) ways of engaging with diverse stakeholders; (4) ways of intervention refinement; (5) identifying key uncertainties and (6) economic considerations.The process evaluation will focus on the first five areas of inquiry, and the overall objectives have been framed to address these key areas.

Study setting
The ARTEMIS cRCT will be implemented in 60 slum clusters across two cities New Delhi and Vijayawada in India.In each city, 30 slum clusters will be included.For this study, a slum cluster is defined as slums within wards or geographical areas identified as slums/resettlement colonies.A ward is a local authority area, typically used for electoral purposes.In certain cities of India, such as Mumbai and New Delhi, a ward is an administrative unit of the city region; a city area is divided into zones, which in turn contains numerous wards.New Delhi is a metropolis and one of the largest cities in India with a population of about 17 million and an estimated slum population of about 2 million. 16 17Vijayawada is one of the largest cities in the state of Andhra Pradesh with an urban population of over 1.0 million 18 19 and an estimated slum population is estimated to be about 0.5 million. 17The most widely spoken language in Delhi is Hindi while in Vijayawada it is Telugu.

Patient and public involvement
Adolescent Expert Advisory Groups (AEAGs) have been formed at each site.This group was involved throughout the intervention development phase in providing inputs and suggestions.Their contributions were critical for the cocreation of the anti-stigma content for adolescents.
During the formative phase, 34 meetings were held where the AEAG provided valuable feedback on ways to engage adolescents in the anti-stigma campaign.

Study design
The study will use a mixed-method, multiple case study design.Each slum cluster will constitute a 'case' for the study.A total of six cases or clusters will be purposively selected taking a maximum variation approach.Slums will be purposively selected to represent different contexts, coverage and reach of intervention as well as the ease or difficulty of implementing the intervention.

Intervention description
The ARTEMIS cRCT will test an intervention to address depression, increased risk of self-harm/suicide or other significant emotional or medically unexplained complaints among adolescents living in urban slums in two cities in India.ARTEMIS has two components.The first is a campaign that aims to reduce stigma related to mental health and improve attitudes and behaviours towards adolescents with depression or at increased risk of self-harm/suicide.The second is a technology-enabled mHealth platform with an integrated electronic decision support system (EDSS), to help primary care doctors and NPHWs to diagnose and treat adolescents at high risk of depression, self-harm or suicide.Before randomisation and the start of the intervention, a team of trained field investigators will screen adolescents at high risk of depression or suicide using Patient Health Questionnaire-9 (PHQ-9), which is a standardised psychometric tool for screening depression and suicide risk in the community. 20 21Adolescents who obtain a PHQ 9 score of ≥10 and/or a score of ≥2 in the suicide risk question on the PHQ-9 will be deemed as 'high risk'.Due to the time delay between screening and randomisation and potential natural remission, a second screening will be carried out for adolescents identified as 'high risk' before the baseline is administered.This process will help identify the final list of adolescents at 'high risk' in all the clusters.The following information will be collected from the study cohort by the trained field investigators: sociodemographic characteristics, history of mental illness, treatment history, comorbid conditions, stressful events experienced in the previous year, resilience, knowledge attitude and behaviours related to mental health and stigma associated with help seeking.A detailed protocol of the trial has been published. 22 theory of change model for the intervention was developed with key stakeholders during the formative phase of the study in 2021.The logic model for intervention has been provided (figure 1).

Intervention arm
In the intervention arm, the two components (1) the anti-stigma campaign and (2) the technology-enabled mHealth-based EDSS for screening, diagnosis and management of depression and self-harm will be delivered.(figure 1).

Anti-stigma campaign
The anti-stigma campaign will include several Information Education and Communication (IEC) materials containing key messages around addressing mental health stigma, myths related to mental health, highlighting major stressors and their impact on mental health of adolescents and the importance of help seeking for mental health problems.The IEC materials will include posters, pamphlets, brochures, videos of persons with lived experience, animation videos, audio dramas, games, magic shows, rallies and street plays.The materials will be cocreated with adolescents from the study sites.AEAGs will be created in both locations consisting of adolescents from the community, with the purpose of involving them in working with the implementation team to cocreate the anti-stigma campaign.Details of the process of forming AEAGs and their engagement in shaping the anti-stigma campaign have been published elsewhere. 23

Peer-educators
Based on findings from the formative phase of the study, peer groups were established comprising adolescents from the community in the age group 13-19 years, some of whom were from the study cohort.Their role will be to provide support to adolescents who approach them to discuss problems and to promote mental health awareness in the community.All the members of the peer groups will be provided basic training by the research team on mental health, safe practices while using social media, addiction and substance use and promoting mental wellbeing through engaging in pleasurable activities such as listening to music, sharing problems with trusted individuals, having friends, exercising and eating healthily.

Technology-enabled EDSS
The EDSS platform has been developed to assist primary care doctors and NPHWs to screen, diagnose and manage depression, other significant emotional or medically unexplained complaints and suicide risk among adolescents.High-risk adolescents will be referred to the doctor electronically via the mHealth platform by the NPHWs.
In each cluster, one doctor and several NPHWs will be trained.The NPHWs will be trained to use the EDSS to provide basic supportive advice to adolescents at high risk of depression or self-harm, refer them to the primary care doctors, and track treatment advice given by the doctors and will follow-up on treatment adherence.The NPHWs will make regular home visits to encourage adolescents at high risk and their families to seek help from the doctor.The NPHWs will also use the EDSS to clinically monitor the severity of high-risk individuals at 4 months and at Open access 10 months after the start of the intervention.High-risk individuals will be provided referral cards to take to the primary care doctors.The primary care doctors will be trained to use an EDSS to clinically diagnose and manage adolescents with depression, other significant emotional or medically unexplained complaints and self-harm/ suicide ideation based on WHO's mhGAP (Mental Health Gap Action Programme)-Intervention Guide algorithm. 24Adolescents needing specialist care will be referred to a psychiatrist in a government facility.The EDSS will also capture and store (on a cloud server) information regarding adolescents' visits and the type of care provided by the primary care doctor-counselling/pharmacotherapy/combination of both/referral/follow-up.All data will be encrypted and protected using userdefined passwords.The information will be visible to the NPHWs, the project supervisors and research team only.A colour-coded traffic light system integrated in the application will assist the NPHWs to identify high-risk adolescents who need to be prioritised for follow-up.The EDSS will also provide them with simple questions tailored to the priority level as indicated by the traffic-light coding system, that they can ask the adolescents during follow-up to ensure appropriate treatment adherence or follow-up with doctors or mental health specialists.Open access

Continued
Open access

Control arm
The control arm will receive enhanced usual care that will include general psychoeducation through pamphlets to raise awareness of common mental disorders in adolescents among the wider community.Parents/guardians of adolescents identified as high risk will be advised to consult a primary care doctor or mental health professional.Those with high scores on PHQ-9 (a score of ≥15 on PHQ9) and/or at high risk of suicide (suicide score ≥2), indicating severe depression and/or high risk of suicide will be asked to seek care immediately and will be given a list of health facilities (with contact details) where such treatment is available.All NPHWs will be asked to follow up with the adolescents and their families more intensively and will recommend that they visit a doctor/ psychiatrist immediately.

Data collection
Quantitative data for the process evaluation will draw on metrics collected and sourced from the open-source

Intervention acceptability and adoption
► What was the experiences of NPHWs in using EDSS for providing care (challenges, perceived benefits, potential for routine use of mHealth)?► What was the experiences of doctors about using EDSS for providing care (challenges, perceived benefits, potential for routine use of mHealth)?► What were some key features of use of EDSS by doctors: -Average time taken for diagnosis and identification of treatment plan using mhGAP over time.-Association between type of CMD and time taken for diagnosis and identification of treatment plan using mhGAP.► What were the perceptions of high-risk adolescents and their parents about ease of getting treatment through mHealth?Open access medical record system or OpenMRS; (OpenMRS.org).
The OpenMRS is a standardised community-driven open-source software for storing and processing medical record information.Quantitative data will be collected throughout the intervention via the mHealth platform.This will include data on screening outcome, treatment seeking, follow-ups and treatment provided for the highrisk cohort.Data will also be used to understand usage patterns of the EDSS by primary care doctors and NPHWs.
The mHealth platform will capture this data in tablets used by NPHWs and primary care doctors for screening, treatment and follow-up of the high-risk cohort.These data will be used to assess reach, effectiveness and service utilisation.Data on competency and fidelity measures will also be captured throughout the intervention.
Qualitative data collection will include key informant interviews (KIIs) and focus group discussions (FGDs) with various stakeholders including high-risk and nonhigh-risk cohorts in both arms, parents, primary care doctors, NPHWs, members of the AEAG and peer leaders.A total of 33 FGDs (each with 6-12 participants) and 28 interviews are planned across the 6 purposively selected slum clusters equally divided between both sites.This will include four intervention clusters (two each in Delhi and Vijayawada) and two control clusters (one each in Delhi Vijayawada) (table 1).The KIIs and FGDs with stakeholders will help capture information related to fidelity of the intervention and identify any gaps that could be addressed.Written informed consent will be sought from all participants selected for KIIs and FGDs.

Data analysis
The conceptual framework for the study and how it integrates with the key parameters from the MRC frameworks is provided in table 2.
The case study methodology will also be used as it is recognised to be suitable to study complex interventions in different contexts and helps to capture 'the complexity of the case, the relationship between the intervention and the context and how the intervention worked (or did not)'. 25Following Pfadenhauer et al, we define context as 'a set of characteristics and circumstances that consist of active and unique factors that surround the implementation.As such it is not a backdrop for implementation but interacts, influences, modifies and facilitates or constrains the intervention and its implementation'. 26e will be selecting both intervention and control sites as cases to avoid the Hawthorne effect.Two slum cluster in the control arm will also be selected for the case study to enable comparison, to understand contextual factors and any changes to usual care in the absence of an intervention.The importance of studying changes in control arms has been recognised as being useful in understanding intervention impact. 27A multiple case study design will be used where each case will be analysed at the case level, but we will also pull together all the cases from the intervention and control arms for a more in-depth analysis and comparison.Basic descriptive analysis will be conducted with the quantitative data.Qualitative data will be transcribed by a professional vendor and the transcripts will be read by at least two members of the research team and coded.Coding will be carried out using a priori codes based on the conceptual framework (table 3) with the help of NVivo 12 software.
Any additional codes emerging during the analysis will be added to the coding framework.Members of the research team will read transcripts separately and come up with codes.The research team will then compare and combine their codes to evaluate their fit and usefulness and will examine the differences in code to see if any new insights can be generated.The code list will be finalised after a discussion between the researchers to establish agreement among coders.Qualitative and quantitative data will be triangulated to arrive at a comprehensive understanding of the intervention.Triangulation helps in confirming the findings from quantitative and qualitative data thereby increasing validity as well as leading to a better understanding of phenomena being studied. 28he quantitative data will be used understand implementation outcomes.Indicators to study implementation outcomes will be informed by the conceptual framework and relevant implementation science theories and frameworks (table 2).1][32] The APEASE criteria have some similarities with RE-AIM but includes two additional parameters, which are-practicability and safety.An intervention is practicable if 'it can be delivered as designed through the means intended to the target population'. 30Data will be extracted from OpenMRS and descriptive statistics will be used to get totals, proportions and changes over time.

Assessment of implementation fidelity
Implementation fidelity refers to the extent to which an intervention was implemented as intended.We will assess fidelity in three components, such as (a) delivery of anti-stigma campaign, (b) implementation of the mHealth component and (3) trainings.Indicators have been developed for all these components to measure the frequency of exposure and coverage.The quality of training provided to primary doctors on the mhGAP, NPHWs on the priority listing app and peer educators on promoting mental well-being and competency will be assessed through various tools including a post-training satisfaction survey, pre-post test self-assessment checklists as well as though rating by trainers and staff on competency checklists.

Discussion
This paper describes the design of a mixed-method process evaluation for the ARTEMIS cRCT, which Open access involves an anti-stigma campaign and a mobile devicebased decision support system for primary care doctors and NPHWs, to improve treatment of adolescents at high risk of depression, other significant emotional or medically unexplained complaints and suicide risk.
The process evaluation will help us understand and explain key causal mechanisms that led to change and will, therefore, strengthen the understanding of the implementation process by highlighting various barriers and facilitators.Furthermore, it will provide an understanding of how the local context played a role in the way the intervention was implemented and help identify the need and impact of any adaptations made to the intervention.The process evaluation will provide stakeholder perspectives on aspects of the intervention that worked and those that need further adaptation.

Strengths and limitations
The protocol was developed using existing implementation science theories and frameworks and combines qualitative and quantitative measures to understand key aspects of how the intervention was implemented and which aspects worked or need further improvement.The MRC framework, which is a comprehensive framework, widely used in the field of implementation science will be used.The case study method using a maximum variation approach will be used to make suitable comparisons among different contexts and avoid Hawthorne.The ARTEMIS intervention does not directly address social determinants which impact mental health outcomes.In this study, we will collect additional qualitative and quantitative data on social support, socioeconomic conditions, education and the local context of participants which will help to better understand and explain linkages between mental health outcomes and social determinants.
Contributors The paper was conceptualised by PKM.AM wrote the original draft with contributions from SKY. PKM commented on multiple drafts before sending a prefinal version to everyone listed as authors.SK, SP, HL, UR, BME, RaS, ReS, DP, RN and GT reviewed the draft and provided critical intellectual inputs and comments to the draft.PKM led the implementation of the trial in India along with SKY, SK, SP and AM.SKY, SK and SP played a key role in implementing research activities in the sites.All authors read and approved the final manuscript.Disclaimer The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care.The funding agency had no role in the writing of the manuscript or the decision to submit it for publication.

Competing interests
We declare that the George Institute for Global Health has a part-owned social enterprise, George Health Enterprises, which has commercial relationships involving digital health innovations.

Type of group/ individual Some areas of inquiry
AEAGs, Adolescent Expert Advisory Groups; EDSS, electronic decision support system.
Table 3 Continued

Figure 1
Figure 1 Logic model for ARTEMIS.IEC, Information Education,Communication; IT, Information Technology; ASHAs, Accredited Social Health Activists.

Funding
Adolescents' Resilience and Treatment nEeds for Mental health in Indian Slums (ARTEMIS) is funded by UK Research and Innovation/Medical Research Council (UKRI/MRC), Grant no: MR/S023224/1.SKY, SK and SP are all partially or fully supported by ARTEMIS.PKM and AM are partially supported through NHMRC/ GACD grant (SMART MentalHealth-APP1143911) and ARTEMIS.PKM is the PI for ARTEMIS and Co-PI for SMART Mental Health.HL and GT are partly support by ARTEMIS.GT is also supported by the National Institute for Health Research (NIHR) Applied Research Collaboration South London at King's College London.GT is also supported by the MRC-UKRI in relation to the Indigo Partner ship (MR/R023697/1) awards.

Table 1
Qualitative data collection plan

Table 2
Conceptual Framework for Process Evaluation