The Scottish Government

« Previous | Contents | Next »

Listen

SUICIDE AND SUICIDAL BEHAVIOUR: ESTABLISHING THE TERRITORY FOR A SERIES OF RESEARCH REVIEWS

CHAPTER TWO: METHODS

2.1 The project was undertaken as a series of linked stages which related to the stated objectives of the scoping exercise. At each stage the research team worked with the Scottish Executive's Mental Health Research Team and the Choose Life National Implementation Support Team to agree the methodological approaches and review findings.

2.2 Stages 1, 2 and 3 comprised the following:

• Stage 1: establishing contact with researchers across the world specialising in suicide and suicidal behaviour research

• Stage 2: conducting a survey identifying research specialists' areas of interest and expertise, and perceptions of the issues and gaps in the evidence base

• Stage3: identifying the focus and structure of the reviews to be commissioned, with input from an expert group

STAGE 1: ESTABLISHING LINKS WITH RESEARCHERS

2.3 As a first step towards identifying the territory for a series of research reviews to collate and co-ordinate the evidence base for suicide and suicidal behaviour, the project set out to draw on the knowledge and experience of research specialists from across the world. In addition, this stage intended to identify and establish contact with researchers in Scotland and beyond who were interested in issues relating to suicide prevention policy in Scotland.

Sample selection

2.4 A sampling frame of research specialists was constructed using 4 methods:

• The mailing list of the British Isles Suicide Research Group.

• Systematic analysis of the Medline database

• Search of peer reviewed articles and references identified in recent unpublished literature reviews of suicide and suicidal behaviour

• Asking researchers identified via the above methods to nominate other researchers with whom they are in contact (snowballing)

2.5 In targeting researchers specialising in suicide-related research across the world, the following criteria were laid out:

• Published outputs must relate directly to suicide and/or DSH

• Publications must have been within the last 5 years (from 1998 onwards) These criteria were set in order to produce a manageable sample size and also a sample of specialists who were likely to have an active and continuing interest in the field of suicide and suicidal behaviour

Making contact and recruitment to the research

2.6 Once identified, research specialists were contacted and introduced to the research. They were informed that the Scottish Executive would be commissioning a number of reviews of previous research on suicide and suicidal behaviour and that the reviews would identify gaps in the existing evidence base, allowing primary research to be commissioned in areas where reliable evidence was lacking. It was made clear that the reviews would focus on risk and protective factors in relation to priority groups, as well as co-ordinating information on prevalence and reviewing evidence on the effectiveness of specific preventive interventions. It was emphasised that the evidence base was intended to inform the planning of national and local actions to address the objectives of the first phase of the Choose Life strategy.

2.7 In order to help respondents (particularly those from outside Scotland) to focus on suicide prevention policy in Scotland, a summary of the key aims and objectives of Choose Life and the key risk priority groups were brought to their attention. A web link to the Choose Life strategy document was also provided.

2.8 Researchers were contacted by email and invited to participate by completing an electronic questionnaire. All electronic contacts had to comply with the provisions of the Data Protection Act 1998 regarding informed consent and disclosure for personal data and the December 2003 Directive on Privacy and Electronic Communications regarding unsolicited email.

2.9 The personal email sent to sample members stated where their contact details were sourced and they were given the option to have their contact details removed from the project's records if they preferred not to be contacted again. In all email correspondence and on the electronic questionnaire, the SDC and the Scottish Executive data privacy statements were provided. Those completing the questionnaire were asked to indicate that they had read and understood these statements.

STAGE 2: IDENTIFYING AREAS OF INTEREST, PRIORITY ISSUES AND PERCEIVED GAPS IN THE EVIDENCE BASE

Data Collection

2.10 A survey of identified research specialists was undertaken using a structured electronic questionnaire. Sample members were contacted by email, as described above, and invited to participate in the survey. The email contained an internet link to the survey questionnaire which was in place on the SDC website for the duration of the survey. Sample members were requested to complete the web-based version of the questionnaire and submit it directly; or to access a Microsoft Word version of the questionnaire available on the SDC website and return it as an email attachment. Those sample members who were unable or unwilling to use electronic communication to submit their questionnaire were sent a paper version of the questionnaire.

2.11 The questionnaire (designed by the research team in consultation with the research commissioners) covered the following:

• Contact details

• Specialist research interests relating to suicide and suicidal behaviour (to a maximum of 5)

• The capacity in which the research specialist worked as a researcher: as a solo researcher, as part of a research team, within a research team or centre based at a university, or within a research team or centre located outside the university sector

• Details of up to 5 of the research specialist's recent or key publications relating to suicide and suicidal behaviour

• What the research specialist considered to be the most constructive and useful approach to dividing up the proposed reviews of previous research on suicide and suicidal behaviour to ensure comprehensive coverage of all issues, while avoiding duplication

• What the research specialist considered to be the main issues upon which the reviews should focus (to a maximum of 5)

• What the research specialist perceived to be the main gaps in the evidence base relevant to the implementation of the Scottish Executive's suicide prevention strategy, with the option of listing up to 5 main issues (which could include primary research, meta-analysis and/or research synthesis)

• Whether the research specialist had links with other leading researchers in the field of suicide and suicidal behaviour, and, if so, what their contact details were

2.12 Research specialists were also asked if they were willing to:

• Have their personal details included on a Scottish Executive database of expert research contacts

• Be contacted by the Scottish Executive in connection with an invitation to tender for one or more of the proposed literature reviews following this research

• Be contacted by the Scottish Executive in connection with an invitation to tender for primary research once the research reviews had made clear where reliable evidence was lacking

2.13 Participants were given 30 days to complete and return the questionnaire. Two weeks after the original invitation emails were sent out, a reminder of the expected completion date was sent to those who had not yet replied; a second reminder was then sent out in advance of the deadline. Access to the electronic questionnaire via the SDC website remained open for one week after the deadline, allowing extra time for late responses to be submitted.

2.14 The questionnaire data submitted directly via the SDC website were entered automatically by participants onto a Microsoft Access database on-line with the necessary verification processes to ensure validity of responses. Emailed and paper responses were input manually to the database by the SDC research team.

Analysis of Survey Data

Development and allocation of data codes for 'interests', 'issues' and 'gaps'

2.15 Codes were applied to the responses relating to the 3 key questions in the survey: the participants' specialist interest areas; the main issues on which they felt that the reviews should focus; and the perceived main gaps in the evidence base relevant to the implementation of Choose Life. For the purposes of this research these 3 questions will be identified as 'interests', 'issues', and 'gaps'.

2.16 In order to work with the large amount and diversity of data supplied, the 3 key questions were coded under a universal set of themes. This enabled the generation of frequencies of codes to assist in the identification of priority topics for the focus of the research reviews. It also supported comparative analysis of the responses across the different key questions.

2.17 A total of 88 themes were derived from the responses. A full list of these coded themes is available in annex 2. Themes were grouped under the following main categories:

• Epidemiology e.g.:

  • International / cross cultural epidemiology

  • Epidemiology of deliberate self-harm

  • Epidemiology (miscellaneous). (This code was used where people gave a very broad response, e.g. 'epidemiology of suicide in the general population' or 'epidemiology'.)

• Determinants e.g.:

  • Neurobiological / genetic

  • Socioeconomic

• Risk factors and conditions e.g.:

  • Alcohol and drugs

  • Physical illness and disabilities

• Protective factors (only one general code was required as protective factors were rarely mentioned)

• Interventions e.g.

  • Crisis intervention

  • Mental health promotion

  • Interventions (miscellaneous). (This code was used where participants had either given a very broad response, e.g. 'interventions' or 'treatments', or had given a very specific response that did not fit into any of the other intervention codes, e.g. 'experimental writing paradigm'.)

• Cross-cutting e.g.

  • Availability of means

  • Media / stigma / contagion / attitudes

2.18 The 'cross-cutting' category included themes such as ' mental illness' and 'ethnic minority/cultural issues' which cut across the other main categories.

2.19 When the themes were sorted under the main categories above they were labelled with code names based on a simple alphabetic system. The first code name was AA, followed by AB, AC to AZ, followed by BA to BZ and so on.

2.20 For some responses to the key questions more than one code was allocated, for example the response 'risk factors - depression' was allocated the following codes:

2.21 If a participant mentioned a theme more than once in a response to a key question, the relevant theme code was allocated only once, in order to prevent skewing of the data. If themes had been coded on every mention, it would result in a participant who mentioned the same interest five times being allocated five interest codes as if they had five interests, whereas in reality they were only describing one interest. The following examples show how the coding process was applied to responses to different questions.

2.22 When asked what their main specialist interest area was, participant 95 listed the following interests:

• General hospital services for people who self-harm

• Involving users in self-harm services

This response was allocated the following codes:

Therefore code BQ 'Interventions - deliberate self-harm' was only allocated for this participant once, even when they had mentioned it twice.

2.23 Similarly when participants were asked what they considered to be the main issues upon which the reviews should focus, participant 75 listed the following issues:

• Risk factors among persons with mood disorders

• Risk factors among persons with schizophrenia

This response was allocated the following codes:

Codes AV 'Risk factors and conditions' and DD 'Mental illness' were only allocated for this participant's responses, even when they were mentioned twice.

2.24 Finally, when participants were asked what they perceived to be the main gaps in the evidence base relevant to the implementation of the Scottish Executive's suicide prevention strategy, participant 68 listed the following gaps:

• Cross-cultural comparison

• Research in Asian countries given the unique epidemiological features

This response was allocated the following code:

Allocation of data codes for participant's suggestions for the most constructive and useful way to divide up the reviews

2.25 The responses to the survey question that asked participants to describe what they thought was the most constructive and useful way to divide up the reviews were summarised and coded using a separate set of themes from the list of themes generated for interests, issues and gaps.

Building a database of interests and expertise

2.26 A Microsoft Access database was produced containing the contact details, publications, and interests of the research specialists who participated in the survey and their responses to the survey questions. This was handed over to the Scottish Executive mental health research team to use as a resource to inform their future commissioning of reviews on suicide and suicidal behaviour and any associated primary research.

Development of a framework for dividing up the reviews

2.27 Frequencies of the codes applied to the responses relating to the interests, issues and gaps questions were calculated. The frequencies were then sorted in descending order to identify the order of priority of themes for each question.

2.28 Frequencies of the coded themes for the responses to the question on the most constructive and useful way to divide up the reviews were calculated. The frequencies were then sorted in descending order to identify the most commonly suggested methods for dividing up reviews.

2.29 The above data was used to develop the key elements of a framework for dividing up the reviews. The framework reflected the priority issues and perceived gaps and most common suggestions for dividing up the reviews, whilst also enabling the inclusion of all issues, perceived gaps and suggestions for dividing up the reviews. The framework then served as a reference tool for stage 3 of the project. Figure 2.1 gives a simplified version of the framework.

STAGE 3: IDENTIFYING THE FOCUS AND STRUCTURE OF REVIEWS TO BE COMMISSIONED

2.30 The key task of the project was to define the scope of the proposed research reviews and to ascertain how best the reviews might be focused and structured. The analysis of survey data assisted with a preliminary identification of priorities for focus and structure.

2.31 The purpose of stage 3 of the project was to assess and filter these priorities in light of the Scottish context in order to generate the territory for the reviews. This involved establishing a group of experts who then worked through a consultative process with the research team's analysis of the survey information to produce an agreed shortlist of priority review topics.

2.32 The membership of the expert group was discussed and agreed between the research team and the commissioners. It was considered essential to balance the research specialist perspectives gained in the survey with policy and implementation perspectives from a more 'home grown' expert group. It was also agreed that, in order to build capacity and develop networks within Scotland as the potential basis for future collaborative research, the expert group should consist mainly of Scottish policy makers and practitioners with relevant research, practice and policy experience, knowledge and contacts.

2.33 A number of key individuals fitting these criteria were then identified and invited by letter to join the expert group by the research team. The expert group included the following perspectives, with some experts representing more than one perspective:

• Service delivery managers and practitioners

• Choose Life local co-ordinators

• Scottish Executive civil servant public health and social policy makers

• Mental health service users

• Those working in suicide prevention services

• Public health consultants

• Psychiatrists

• Scottish Prison Service

• Local health and social services commissioners

• Academic experts from Scotland and other parts of the UK

• Mental health government research commissioner

A full list of those who participated as members of the expert group is available in annex 1.

Consultation with expert group: phase 1

Purpose

2.34 The research team used an adaptation of the classic Delphi technique for the purposes of their consultation with the expert group.

2.35 The aim of the Delphi technique is to elicit information and judgments from expert participants to facilitate problem solving, planning and decision making. It does so without physically assembling the participants. Instead, information is exchanged through mail, fax or email. The process consists of a variable number of phases, including: identification of the issue and solicitation of ideas, collation of responses and request for refinement and elaboration from participants, further collation of responses, further request for refinement until resolution through consensus is achieved.

2.36 The Delphi process was adapted for the purposes of this project to enable 2 consecutive consultation rounds with the expert group. The consultations aimed to gain affirmation from the expert group that the analysis of the survey data was sound and that the suggested review topics reflected a focus and structure compatible with their opinion of how the reviews could usefully contribute to the implementation and development of suicide prevention policy in Scotland. The consultations were conducted with a view to arriving at an agreed shortlist of research review topics that:

• Met the specific needs of Scotland in relation to the implementation of Choose Life

• Covered all aspects of suicide and suicidal behaviour whilst avoiding over duplication

2.37 Both consultation rounds were conducted without assembling the participants. The first consultation round was to assess and refine the findings of the survey; the detail of how this was done is given below.

Data collection

2.38 Each member of the expert group was sent a guide to the procedures for the 'Delphi' process. A summarised version of the document sent to the expert group ('Delphi Process: Guide to Phase 1 for Expert Group') can be found in annex 3. The guide contained the following information:

• Description of the research specialists who participated in the stage 2 survey

• A first analysis of the data gathered in the survey and a copy of the framework for dividing up the reviews; coding was used to show how the main issues for the reviews and gaps in research suggested by participants fitted into the elements of the framework for dividing up the reviews

• A copy of the original survey questionnaire

• A set of questions to respond to within 3 weeks

This information was given to the expert group to provide them with an opportunity to become familiar with the original data and the process of analysis that had taken place so far.

2.39 Members of the expert group were sent a covering letter asking them to read the information in the guide and to answer the following questions set by the research team:

• Does the framework for dividing up the reviews make sense of the data?

• What is missing?

• Are the issues and gaps matched well with the components of the framework?

• Which matches are missing or need to be changed?

• Comment on methodological considerations

• Comment on questions that the reviews should answer

Data analysis

2.40 The research team then used the responses from the expert group to the questions above to establish a robust starting point for the generation of research review topic options. The research team used the feedback from the expert group to:

• Refine the framework for dividing up the reviews

• Code the survey data to identify the highest priority issues for the focus of the reviews and the highest priority gaps and how they fit into the framework

• Identify how issues and gaps cut across different components in the framework, and the common links

• Assess how often issues and gaps matched components in the framework at different levels to identify priority areas for review

• Decide how the framework should be split into review topics and generate the content of these review topics, making sure that every part of the framework was included within the suggested topics for review

Consultation with Expert Group: phase 2

Data collection

2.41 After processing comments from phase 1 of the consultation with the expert group, the research team produced a set of suggested topics for the reviews. The expert group were contacted and asked to comment on these suggested review topics to help shape the final structure and scope of the research reviews.

2.42 The letter to the expert group emphasised that the research team was particularly interested in their thoughts on how well the suggested topics for the reviews would inform the policy development and implementation phases of Choose Life. The suggested topics for reviews document contained an introductory section which explained to the expert group that all of the reviews would need to stress:

• The importance of including recommendations ('so what?') for policy and practice

• A discussion of the validity of findings, particularly with reference to their transferability to the Scottish context (culture, society, service structures, economy)

• A discussion of the implications of findings for future monitoring, evaluation and research, with particular reference to the Scottish context

• Stipulation that the review research team includes a practitioner and/or a service user (or else can demonstrate that practitioner/user views have been canvassed in the course of producing the review)

• Stipulation that the review research team is a partnership combining diverse disciplinary perspectives/understandings

• The importance of differentiating between deliberate self-harm, suicidal behaviour and suicide

2.43 Participants were asked to structure their comments by reading the suggested review topics and answering the following questions:

For each review topic:

1. Are the most important issues relevant to each topic area given the prominence/priority they deserve?

If no,

1a. What are the issues?
1b. Why are they not prominent?
1c. Why should they have more prominence?
1d. How could this be done?

2. Do the reviews sufficiently address the 'so what?' question to produce maximum understanding and utility in the Scottish context.

If no,

2a. How can the reviews take more account of and be more applicable to the Scottish context?

3. Are the methods suggested appropriate to the task?

If not,

3a. What needs to be added or changed?

For all of the review topics taken together:

4. Do the review topics together sufficiently address the relevant issues and gaps in knowledge concerning:

4a. Suicide?
4b. Suicidal behaviour?
4c. Deliberate self-harm?

If not,

4d. What is missing?
4e. Why should it be included?
4f. How could this be done?

5. Can the review topics be arranged in order of priority?

5a. If no, why not?
5b. If yes, please describe the order of priority you feel would be most appropriate.

Other questions that were specific to individual review topics were included within the text of the suggested review topics in italics.

2.44 Participants were asked to spend no more than an hour or 2 on this task and to respond within 3 weeks.

Data analysis

2.45 The feedback from the expert group to the above questions was used to prioritise and refine the review options. The research team then produced a prioritised list of territories for a series of research reviews. This list was intended to give the Choose Life National Implementation Support Team guidance on how they might prioritise and allocate resources to potential review options, while providing reassurance that all perceived issues and gaps in the evidence on suicide and suicidal behaviour would be covered.

« Previous | Contents | Next »