This study had two phases: a literature search and questionnaire development, and the Delphi process. Please see Figure 1 for a summary of the steps.
Literature search
The aim of the literature search was to find statements which instruct the reader on how to determine whether someone is having thoughts of suicide, how to offer assistance in the short term, and how to access appropriate professional help for a suicidal person. The literature search was conducted across three domains: the medical and research literature, the content of existing suicide prevention and intervention programs, and lay literature. The lay literature included books written for the general public, particularly carers' guides, websites and pamphlets.
The medical and research literature was accessed through searches of PsycInfo and PubMed. The search term was 'suicide' and all records for the 20 years leading to the search date were reviewed. The search term 'suicide' generated far too many records, but all attempts to narrow the search were found to be unsatisfactory as they excluded too many relevant records. Papers which described assessment of suicide risk, brief suicide interventions, or guidelines for treating suicidal patients were reviewed, a total of 234 papers. While much of the advice given in these papers was considered too clinically orientated to be useful for first aid, a number of papers did include brief advice and simple intervention instructions. Statements were drawn from 42 of the 234 relevant records. All statements felt to be simple enough for lay people to use were included.
To find appropriate websites, we used the search engines Google [8], Altavista [9], and Yahoo [10] using the search term 'suicide'; the first 50 websites listed by each were reviewed; beyond the first 50 websites, quality declined rapidly. Since most websites were listed by more than one search engine, only 68 websites were reviewed. The websites were read thoroughly, once again looking for statements which suggested a potential first aid action (what the first aider should do) or relevant awareness statement (what the first aider should know). Any external links to other websites were followed and the same process applied to each of them.
The fifty most popular books on the Amazon [11] website which listed the word 'suicide' in the title or keywords were selected. This site was chosen because of its extensive coverage of books in and out of print, including works about mental health aimed at the public. Books which were autobiographical in nature, self-help guides and clinical manuals were excluded. The remaining books were read to find useful statements. The majority of these were carers' guides, which do contain advice relevant for first aid.
Any relevant pamphlets were sought and read, and statements were taken from these as well. The majority of the pamphlets were written and distributed by organisations focussing on mental health in general, or suicide in particular, but some were more general community organisations. Most of these pamphlets were obtained from websites, but where these were not available online, a request was made for relevant materials from large mental health and community organisations. While the majority of the lay literature focussed on understanding suicide, supporting people who have lost a loved one to suicide, and being aware of the risk factors and warning signs which might indicate that someone was thinking about suicide, there was also some advice to friends and family members on what to do if they are concerned that someone they love may be at risk.
Where available, the training materials from existing courses which address suicide intervention were also reviewed and statements were drawn from these. Only a small number of training courses were found to be relevant, as the majority of such training is developed for professionals with previous clinical training in specific settings. The courses for which material was reviewed were the existing Mental Health First Aid Program [12], the Applied Suicide Intervention Skills Training (ASIST) Program (LivingWorks Canada) [2, 13], and the Mental Illness First Aid Course (Canadian Mental Health Association) [14].
Questionnaire development
The questionnaire was developed by first grouping statements into categories: identification of suicide risk, assessing seriousness of the suicide risk, initial assistance, talking with a suicidal person, no-suicide contracts, ensuring safety, confidentiality, and passing time during the crisis. Similar or near-identical statements were frequently derived from multiple sources, and they were not repeated in the questionnaire. A working group comprised of the authors of this paper and colleagues working on similar projects convened at each stage of the process to discuss each item in the questionnaire. The role of the working group was to ensure that the questionnaire did not include ambiguity, repetition, items containing more than one idea or other problems which might impede comprehension. The wording was carefully designed to be as clear, unambiguous and action-oriented as possible. For example, 'the first aider should find out if the person is thinking about harming themselves' is better stated 'the first aider should ask the person if they have been having thoughts of suicide'. All statements were written as an instruction as shown in the above example. The only items which were not included in the questionnaire were those which were so ambiguous that the working party was not able to agree on the meaning of the statement, or those which called upon 'intuition' or 'common sense', as these cannot be taught.
The majority of participants answered the questionnaire via the Internet, using an online survey website, Surveymaker [15]. Three participants requested paper copies of the questionnaire, as they did not have convenient access to the Internet. Participants were able to stop filling in their questionnaires at any time and log back in to continue, without the risk of losing the completed section of their questionnaires. Using the Internet also made it very easy for the researchers to identify those who were late in completing questionnaires and send reminders, with no need to send extra copies of the questionnaire. No questions were inadvertently missed, as the web survey was set up so that each question was mandatory. In addition, such survey software allows for branching, so participants who did not endorse the use of no-suicide contracts were not asked to answer questions about what such a contract should contain.
Statement selection
The criteria for item inclusion in the questionnaire have been articulated above; items which are non-clinical in nature, interpretable by the research team, teachable and useful to a member of the public with only minimal training were included. To clarify, consider the examples below.
"Consider a brief hospitalisation for your suicidal patient."
While a professional such as a family doctor or psychiatrist may recommend hospitalisation, a member of the public could not take this action themselves. This item is very clinical in nature, and not useful to a member of the public, so was not included.
"You need to walk the walk with the person you are helping."
This statement cannot be interpreted literally, so was not included.
"It is important to follow your instincts when helping a suicidal person."
This statement asks the reader to draw on instinct, which cannot be taught or effectively described. It was not included in the questionnaire.
"If you suspect that someone may be suicidal, you should ask them directly."
This item is clear and concise, will lead to accurate identification of the problem, and can be done by anyone, so it was included in the questionnaire.
The Delphi process
Participants were recruited into one of three panels: professionals (clinicians and researchers), consumers (people who had experienced suicidal ideation or a suicide attempt in the past) and carers. The professional panel had 22 experts, the consumer panel 10, and the carer panel 6. All panel members were from developed English speaking countries (Australia, New Zealand, The United States, England and Canada). Participants were recruited in a number of ways. Professionals recruited were those who had publications in the areas of suicide intervention or prevention, identification of suicidal ideation, or treatment of suicidal patients. When letters were sent to professionals asking them to be involved, they were also invited to nominate any colleagues who they felt would be appropriate panel members. Those active in clinical practice were also asked to consider any former patients who might be willing to be involved. The 22 professional participants included 5 psychologists, 5 psychiatrists, 3 managers of mental health services, 2 social workers, 1 nurse, 9 researchers and 3 professors of psychology. Some participants had multiple roles in research, teaching and clinical work. Consumers were recruited from advocacy organisations, and referral by clinicians. They were also identified if they had written websites offering support and information to other consumers, or published memoirs. Carers were recruited through carers' organisations, but were difficult to recruit for this study. It may be that few carers see themselves as being adequately experienced in dealing with suicide crises, having been involved in perhaps only one. In some cases they may not be aware that the person they care for has been suicidal or even made a suicide attempt in the past. We discussed approaching support groups for people who had lost a loved one to suicide, but it was decided this would be inappropriate and may be distressing to the people in the groups.
Three rounds of questionnaires were distributed as follows, with each statement being rated up to two times. In round 1 the questionnaire, derived from the process described above, was given to the panel members. The questionnaire included space after each of the sections to add any suggestions for new statements that panel members felt should be included.
In each round of the study, the usefulness of each statement for inclusion in the mental health first aid guidelines was rated as essential, important, don't know or depends, unimportant, or should not be included. The options don't know and depends were collapsed into one point on the scale because operationally, they are the same response; most of the statements were, very reasonably, noted to be useful in some cases and not others, meaning they could not be generalised in guidelines, which is also true of statements participants did not feel confident to rate.
The suggestions made by the panel members in the first round were reviewed by the working group and used to construct new items for the second round. Although the carers' ratings were not used in the final analysis, their suggestions in round 1 for new ideas were included in round 2. Suggestions were accepted and added to round 2 if they represented a truly new idea, could be interpreted unambiguously by the working group, and were actions. Suggestions were rejected if they were near-duplicates of items in the questionnaire, if they were too specific (for example, "I get my husband to do some woodwork"), too general ("just be there"), or were more appropriate to therapy than first aid ("develop a strategy for coping with intense emotions, specific to the emotion; depression, anger, guilt").
Items rated as essential or important by 80% or more the professional and consumer panels were accepted for inclusion in the guidelines. If they were endorsed by 80% or more of one of the panels, or by 70–80% of both panels, they were re-rated in the subsequent round. Items which met neither condition were rejected. Before the second and third rounds of the study, each participant was sent a summary of the results of the previous round, listing which items had been accepted, which had been rejected, and which were to be re-rated. When an item was to be re-rated by the panellists, they were provided with their own response and a table outlining how many people in each group had endorsed the item. They were told that they did not have to change their responses when re-rating an item, but that if they wished to, they would have the opportunity to do so.