The study involves a non-controlled retrospective cohort of 1,783 relatives of missing persons in Sri Lanka, who took part in the ICRC’s Accompaniment Programme between April 2016 and August 2017. The data were not initially collected for research purposes but rather as part of the information collected routinely throughout the programme for follow-up and internal monitoring.
Intervention: The MHPSS component of the ICRC’s Accompaniment Programme
Several ICRC departments are involved in addressing the multi-faceted needs of the families of missing persons. The specific mental health and psychosocial support (MHPSS) component is carried out in collaboration with local partner NGOs and selected relatives of missing persons trained to work as accompaniers. All accompaniers are female in order to avoid sending males into female-headed households. They are trained and supported by a multidisciplinary team to understand and address the wide range of needs of the families of missing persons and to provide a bridge to local services.[i] The MHPSS intervention consists of a six-month cycle during which various activities are carried out (table 1).[ii]
The literature on ambiguous loss purports that the emotional difficulties and manifestations of common mental health problems like anxiety and depression are not to be seen as psychiatric disorders, but rather as a social and relational problem.[i] Thus, the therapeutic goal is to reinvest in emotional attachment, garner social support and build resilience to cope with the ambiguity. The intervention guidelines for working with ambiguous loss[ii] were integrated as culturally and contextually adapted themes into eight support-group sessions offered to relatives of missing persons. In addition to conducting group sessions, the ICRC-trained accompaniers also carried out individual home visits and referred families with particular needs (financial, legal, medical or other) to local service-providers under the joint supervision of a coordinator from the partner NGO and an ICRC psychologist.
The accompaniers were selected from the area in which the target families lived and, in most cases, they were themselves relatives of missing persons. The minimum criteria for selection included: high school education, emotional stability (assessed through HADS), mobility (i.e. able to travel around), female gender (to avoid sending males to female-headed households) and motivated to support other relatives of missing persons (assessed through selection interview).
In order to build the capacities of the accompaniers using adult participatory methods, the initial five-day training session included basic support skills (empathic listening and interviewing techniques), how to conduct assessments (i.e. how to use the self-reported questionnaires to assess multi-faceted needs, and the self-reported psychological distress questionnaire), resource mapping (to understand all the resources linked to the different needs of families), understanding and managing stress, and basic techniques of problem-solving-based counselling. Immediately after the initial training session, the accompaniers went to the field to put these skills into practice, under the supervision of the coordinators and the ICRC psychologist. This was followed by a second four-day training session on how to conduct support groups. The training was designed in such a way that the theme-centred group discussions were experienced by the accompaniers first, which allowed them to get first-hand experience of the content of the sessions. To ensure that there was no power dynamic between the accompaniers and the families they supported, the ICRC was in charge of identifying the families to be supported, and the ICRC MHPSS team carried out regular supervision sessions to ensure that the multifaceted needs of all of the families were addressed.
The peer groups consist of eight sessions designed as follows: Session I deals with introductions, setting group norms including confidentiality, constructing a group identity with a group name, and understanding and clarifying the objectives of the support groups. Session II introduces the concept of family system and focuses on helping family members revisit and value the different roles and functions that they have in their lives, while being aware of the roles and functions of other family members. The session aims to facilitate a re-evaluation of the relative’s identity, and thus paves the way for reconstruction. Session III introduces the concept of ambiguous loss and its effect on our mind, our body, our behaviour and our relationships. The session aims to recognize the source of the problems derived from having a missing relative within the social and political context and to normalize the reactions that relatives experience. The session also includes a discussion of the various coping mechanisms that each relative uses within his or her social and cultural context.
Session IV shifts focus away from the relative towards revisiting the memory of the missing person. Each participant shares a personal positive remembrance with the group. At the end of the session, the group members are encouraged to bring something in memory of the missing person (such as an object, food, song, etc.) to the following session to share with the group if they wish to do so. In sessions V and VI, group members are encouraged to discuss ways of paying tribute to their missing relatives as a group. Group members brainstorm ideas and come up with a commemoration event. The commemoration event seeks to allow group members to carry out a meaningful ritual or tribute and gain recognition for their missing relatives, and for themselves. For the actual commemoration event, families invite other community members, including village and/or religious leaders, as well as representatives of authorities where appropriate.
Session VII offers a reflection on the social support network of the group members. This session aims to help them realize that they are not alone and that there are people who are there to help them. A personal map is drawn as a tool for this exercise. Finally, session VIII is a joint evaluation of the group process. The group members are also encouraged to discuss whether they want to continue to meet on their own, and how they would see this through.
Sources of data
Data concerning the missing persons whose relatives participated in the Accompaniment Programme
- Tracing request: Collected when the relatives approached the ICRC between 1989 and 2009 to register their missing person. The data were collected by an ICRC protection staff member using a closed questionnaire in Sinhala or Tamil when the relative(s) approached the ICRC for registration. The staff member immediately translated and recorded all the information in English in the ICRC Protection database. The following data were used in this study:
- Year of disappearance
- Age at disappearance
- Occupation (combatant or civilian status)
- Region (north, east or south)
Data concerning the relatives who participated in the Accompaniment Programme
- Household survey: Collected in 2016 and 2017 when launching the MHPSS activities in a given area, conducted in Sinhala or Tamil by the ICRC-trained accompaniers. Qualitative parts of the household survey were translated into English and entered into the ICRC Protection database by an ICRC data-entry operator. Data used in the study:
- Relationship to the missing person
- Financial status as estimated by the accompanier (extremely poor, poor, moderate, well-off)
- Region (north, east or south)
- MHPSS pre-, post- and retention assessments: Conducted with all 1,783 relatives in 2016 and 2017 prior to the MHPSS interventions (pre-assessment), upon completion six months later (post-assessment) and additionally for 285 relatives approximately one year after the activities had ended (retention assessment). All of the MHPSS assessments are identical and include the 14-item hospital anxiety and depression scale (HADS),[iii],[iv] a seven-item, shortened version of the Bradford Somatic Inventory (BSI shortened) as well as a seven-item functioning scale developed by the ICRC using the free listing method. The data were collected in Sinhala or Tamil and, as they were numerical, there was no need for translation prior to entry into Excel. Data used in the study:
- Pre, post and retention HADS scores for anxiety (range 0-7: low: 8-10: mild; 11-14: moderate; and 15-21: severe)
- Pre, post and retention HADS scores for depression
- BSI shortened pre, post and retention scores for somatic difficulties (range 0-7: low; 8-10: mild; 11-14: moderate; and 15-21: severe)
- ICRC functioning scale pre, post and retention scores for daily functioning (rated 0-32 with no standardized cut-off scores) (see scale description in the next section).
Data concerning the MHPSS intervention
- Group session attendance sheets: At each of the eight peer-support sessions offered to relatives, attendance sheets were collected by the accompanier and entered into the ICRC Protection database by a data-entry operator. As the data were numerical, there was no need for translation. Data used in the study:
- Attendance of sessions one to eight
- Individual home visit sheets: After each individual home visit, attendance sheets were collected by the accompanier and entered into the ICRC Protection database by a data-entry operator. Data used in the study:
- Total number of home visits received
- Referral sheets: Referrals of relatives to local service-providers were recorded by the accompanier and entered into the ICRC Protection database by a data-entry operator. Data used in the study:
- Referred to one or more local service-providers (yes/no)
Locally developed functioning scale
Given that no scale for measuring daily functioning in the Sri Lankan context currently exists, the ICRC decided to create a local functioning scale using the ‘free listing’ method (Poudyal et al, 2009;[v] Bolton et al, 2002[vi]). This method consisted of asking 20 adult male and 20 adult female members of families of missing persons, from both the language groups, Sinhala and Tamil, what a “functioning person” in their community would have to be able to do for him- or herself and for others. The sample included a balanced number of both young adults, aged between 21 to 40, and older adults, aged 41 to 60, for both men and women.
The responses from the free listing were compiled in the local languages as a core list and were categorized into the themes that emerged. For example, putting on make-up, cleaning yourself, getting dressed and shaving were all grouped under the theme of “taking care of yourself” with one composite gender-neutral question per theme. A total of seven questions (table 2) were retained based on the frequency of responses, and the themes they could be grouped under. The questions were put into a five-point Likert scale, from “not at all =0” to “too much=4” and piloted for comprehension level in both Tamil and Sinhala before the survey was rolled out. The survey results were used to assess the questionnaires’ internal reliability (Cronbach’s alpha). For the Tamil version, the Cronbach’s alpha was 0.87 (n=235), and for the Sinhala version, it was 0.76 (n=127). This local functioning tool has been used within the framework of the Accompaniment Programme for the outcome evaluation of daily functioning.
Data management and statistical analysis
All data were numerically coded (categorical data). Quantitative data (i.e. scores, age of missing person, age of disappearance) were kept as such and grouped depending on the type of analysis: either by identifying the median (such as improvement in the various symptoms) and quartiles, or by using established clinical cut-offs (e.g. on the HADS scale from 0 to 21, 15 to 21 indicate severe symptoms of depression).
The dataset was created in Microsoft Excel with two independent data clerks to control for potential typing mistakes. The electronic dataset was protected by a password, which was changed every three months. The dataset was transferred to STATA, version 11, for analysis.
All quantitative variables were explored by defining their means (and standard deviation), medians and quartiles. Comparisons of means were tested through the t-test, and the corresponding p-value was reported; 95% confidence intervals (95% CI) were calculated around means and means differences. Categorical variables were explored through percentages and tested using the Chi2 test to retrieve the corresponding p-value; 95% CIs were calculated around these percentages.
To measure associations between variables (crude and multivariable), logistic regression was used to calculate odds ratios (OR) with corresponding 95% CIs and p-values from the Wald test. All variables were initially explored in a crude model and were only included in a multivariable model if they were found significant in the crude analysis. Where results from the crude analysis were very similar (within a 10% variation) to the adjusted ones, only the results of the crude analysis were reported to simplify the analysis.