The study will employ a qualitative design involving in-depth interviews with multiple informants in order to build up a series of detailed case studies of completed suicides.
Sample
In order to build on previous work [27–30], we are retaining our focus on individuals who had no contact with specialist mental health services during the 12 months prior to death. This population accounts for more than 75% of all suicides [31] but remains seriously under-researched. Members of this population pose the greatest challenge in terms of identification of risk and opportunity for intervention. They are effectively hidden within their communities and measures to identify and support them will depend heavily on lay involvement.
Maximum policy relevance will be achieved by focusing on the 18–34 age group. In the UK, the suicide rate among men under the age of 35 is well above that of the general population and causes particular concern as deaths among young people represent the greatest numbers of years of life lost. The study will include both sexes, in order to explore the comparative size and strength of their social networks, the flow of information around the network and other issues relating to prevention.
Inclusion criteria
Individuals whose deaths were recorded as suicides by participating coroners, who died aged 18–34 and who had no contact with specialist mental health services during the 12 months prior to death.
Exclusion criteria
Individuals whose deaths were not recorded as suicides; suicides under the age of 18 and aged 35 and over, and those in which the deceased had received care from specialist mental health services during the 12 months prior to death. Although it is common practice to treat open verdicts as suicides for research purposes, they are excluded from this study as it would be highly problematic to assume that relatives and friends believe the death to be a suicide and to question them as if it were. Even where deaths have been officially recorded as suicides, interviewers need to be sensitive to the possibility that some informants will not regard this as the correct verdict.
The study will produce detailed analyses of 15–20 cases, with as many informants as possible per case, in order to build up a full and nuanced picture of the deceased's social situation. For each case, we are aiming to interview a minimum of 3 people, including a family member, a close friend or peer and a work colleague, as well as the general practitioner (GP). Although not part of the lay network, GPs often have considerable insight into the deceased's family and social situation and may be able to identify ways in which lay people and health professionals could have worked together to support the distressed individual.
Identification and recruitment of informants
Cases of completed suicide are being identified through Her Majesty's Coroners in London, South West England and South Wales. In the UK, in order to comply with the Data Protection Act 1998, researchers may be required to obtain honorary contracts in order to work in the offices of participating coroners and view personal data relating to the deceased.
The process of identifying and recruiting informants is outlined in Figure 1. From coroners' registers, researchers identify all suicide verdicts given within the last 12 months and assess them against the eligibility criteria. They then work consecutively through all eligible cases, examining the coroners' files and entering basic demographic data and salient details relating to family and social contacts directly into a database. Information to be gathered at this stage includes:
This information is stored either on a password-protected desktop computer or on an encrypted CD and remains in the coroner's office until such time as all next-of-kin have indicated whether or not they are willing to participate in the study.
The next-of-kin of all eligible cases are approached in writing by the coroner who conducted the inquest. Decisions as to how soon after the inquest the letter should be sent are taken by coroners and their officers and are guided by their knowledge of each family's circumstances. This is likely to be between 3 and 12 months after completion of the inquest. Care is taken to avoid contacting the family around the time of the deceased's birthday and in the weeks leading up to the first anniversary of the death, in accordance with recommended procedure for psychological autopsy-type studies [20]. Enclosed with the letter of invitation is an information leaflet, a reply slip on which they can indicate whether or not they are willing to consider participating and a pre-paid envelope addressed to the research team.
If a next-of-kin does not wish to participate in the study, his/her personal details and those of all other potential informants, including the GP, are removed from the database. The deceased's case record is anonymised, but basic demographic details remain on the database to allow us to determine the extent to which cases included in the study are representative of the total population. Likewise, if an eligible case has no recorded next-of-kin and neither the information in the file nor conversations with the coroner's officer suggest any potential informants, the deceased's demographic data are entered but the case is logged as one that cannot be investigated. It is important to determine the proportion of such cases, since social isolation and lack of a personal network are known risk factors for suicide [32]. The database is fully anonymised before being removed from the coroner's office for analysis. The only personal details retained by the research team are those supplied by consenting next-of-kin.
On receipt of a reply slip from a next-of-kin indicating willingness to consider participating, a researcher makes contact by telephone, discusses the study and answers any questions. If the next-of-kin wishes to participate, the researcher makes arrangements for the interview to take place, ensuring that the informant knows how to contact the research team, should s/he change his/her mind and wish to cancel the appointment. If the informant requires more time to consider whether or not to participate, the researcher seeks permission to phone again after an agreed interval. Informants are asked to sign the formal consent form immediately prior to commencement of the interview.
The intention is to interview as many members of each deceased person's social network as possible, in order to gain different perspectives on the events leading up to the suicide and the social context in which it occurred. From previous work, it is clear that the accounts that family members give are motivated by a number of factors, including a desire to 'set the record straight' and to assign blame to other members of the deceased's social network [28]. In the present study, we purposively seek out those who may offer an alternative interpretation of events, both in pursuit of 'fair dealing' [33] and in order to build up a balanced understanding of the deceased's social situation and explore any ties or tensions that might have acted as barriers to awareness or impeded efforts to intervene.
As the first interview proceeds and the 'story' unfolds, key players will emerge. At the end of the interview, the researcher seeks the informant's permission to contact further members of the network and to interview the deceased's GP. A contact sheet is used to record the personal details of potential informants and to obtain the next-of-kin's consent to approach them. Nominated individuals are contacted by letter, telephone or e-mail, as recommended by the primary informant.
As each subsequent interview proceeds, the informant is asked if there are any further members of the social network who may have additional insights into the events leading up to the suicide and how it could have been prevented. This technique is commonly known as 'snowball sampling'. Snowball sampling is sometimes criticised on the grounds that it limits the pool to members of a specific network [34]. In certain types of study, however, that is precisely its strength. It has proved to be highly effective in studies of vulnerable, stigmatised, deviant or closed social groups, such as drug takers, homeless people or those affected by HIV/AIDS, as it enables in-depth knowledge of group beliefs and behaviours to be built up and gaps in knowledge to be identified and filled [35, 36]. The researcher uses the trust that is built up in each interview in order to penetrate deeper into the social group and, by means of peer referral, obtain access to a range of perspectives that would otherwise remain hidden or inaccessible.
We continue snowballing until either no further informants are forthcoming or saturation is reached and no fresh information on the case is emerging. The use of progressive snowballing, or chain referral, ensures that our sample in each case is not restricted to those who are sanctioned by the next-of-kin. This is particularly important in the case of young people who may have been involved in drugs or other activities of which their parents were unaware.
Interview methodology
The study uses an in-depth, narrative approach to interviewing that recognises and capitalises on participants' tendency to tell stories, particularly when talking about very painful experiences [28]. This approach relies on a single opening question that is designed to elicit an extended and uninterrupted narrative and to give the participant maximum control over the way in which material is organised and presented [37–39]. The rationale for adopting this approach is both scientific and ethical. By leaving participants free to decide what to include and how to structure their story, rather than asking predefined questions and imposing a structure, it provides far greater insight into the meanings and relative significance that particular events have for them. From an ethical point of view, given the very distressing nature of the subject matter, it also allows the participant to feel more relaxed and secure, knowing that they can disclose information at their own pace [37, 40, 41].
In this study, the participant is asked: 'Please tell me, in your own time, about [the deceased], about your relationship with him/her and about what happened in the period leading up to his/her death.' The interviewer then takes a back seat and assumes the role of 'active listener', noting down points to follow up later but not interrupting. Once the free flow of narrative has come to a halt, the interviewer asks follow-up questions based on the narrative itself, picking up any points within the story that require clarification or elaboration and also offering 'pertinent ways of conceptualizing issues and making connections' for the participant to consider [42]. This second part of the interview is very much a dialogue, in which interviewer and interviewee work together to construct an account of the suicide and the context in which it occurred that addresses the needs of the study. A topic guide is used as a prompt to the interviewer, in order to ensure that all areas of interest are covered.
Finally, the participant is encouraged to construct a visual representation of the deceased's social network, using pebbles, sticky notes and coloured pens to indicate members of the network, their relative proximity to the deceased, the quality of their relationships with each other and the paths by which information travelled around the network. Interviews are audio-taped and transcribed verbatim.
Data analysis
Data analysis begins as soon as the first interview has been conducted and occurs alongside data collection, to allow the interviews to become progressively focused and emerging hypotheses to be tested and refined.
The data lend themselves to three types of analysis: narrative analysis of individual accounts; detailed case studies, built up using the multiple accounts relating to each case, and general cross-case comparisons based on both thematic and narrative analysis. We start with the analysis of individual narratives. On the basis of an initial reading of a number of transcripts by all team members, key themes are identified and a coding frame is developed, which is then applied systematically to all transcripts. As each transcript is coded, the textual material relating to each theme is summarised and recorded, together with line references, illustrative quotations and interpretive comments, on an individual narrative proforma. The proforma allows the overall integrity of each separate account to be preserved and provides the basis for narrative analysis, with its emphasis on structure, plot and performance, rather than on content alone [43]. Participants are offered the opportunity to peruse their own interview transcript and the narrative proforma, in order to validate or contest our interpretation of their story and to offer any further thoughts that have occurred to them since the interview.
We then 'unpack' each theme, breaking the material down further into subtopics and smaller categories and using thematic charts to provide a 'viewing platform' [44] from which to compare different accounts, both within and across cases. The charts are used to produce descriptive accounts of each theme, identify deviant accounts and generate tentative hypotheses for testing against the full data set.
Individual case studies are built up using all the accounts relating to a case, identifying discrepancies and areas of agreement between informants and relating these to their position within the social network. The visual 'map' of the social network that is constructed with participants during the interview is used to contextualise individual accounts. Using a timeline approach, we aim to build up a detailed picture of:
-
what signs of distress each member of the network picked up;
-
how they interpreted these, both at the time and in retrospect;
-
what actions they took, including communicating concerns to other network members and seeking help from outside agencies;
-
what barriers to awareness and intervention operated;
-
what further potential for intervention may have existed;
-
what knowledge and skills might have enabled them to manage the situation differently and possibly prevent a fatal outcome.
The intention is to gain insight into the suicidal process, as it unfolded within a particular micro-social context and as witnessed by social network members. We are particularly interested in the way in which the deceased interacted with and managed his/her social network and the way in which personal relationships may have impacted on the ability to recognise and respond to risk of suicide.
We then look across cases, using the emergent themes to transcend the particular, compare and contrast cases of suicide and develop generalisations that represent the total data set [45]. These will provide the basis for the development of hypotheses and interventions for testing in future studies.