The Delphi process
The general features of the Delphi methodology have been described in the literature . There are many variants, but all involve a group of experts making private ratings of agreement with a series of statements, feedback to the group of a statistical summary of the ratings, and then another round of rating. Delphi group members do not meet, so it is possible to do studies using mail or the internet. The output from the process is statements for which there is substantial consensus in ratings.
Panel members were drawn from the graduates of the International Mental Health Leadership Program . This is an initiative of The Centre for International Mental Health, University of Melbourne, and The Department of Social Medicine, Harvard Medical School. It is a leadership training program and an international network of mental health professionals committed to mental health system development. Questionnaires were sent to 59 mental health clinicians who had trained in the program. Responses were received from 28 from the following countries and territories: Cambodia, (n = 1), China (n = 5), Hong Kong (n = 1), Indonesia (n = 5), Japan (n = 3), Malaysia (n = 1), Mongolia (n = 1), Sri Lanka (n = 1), South Korea (n = 3), Taiwan (n = 1), Thailand (n = 2) and Vietnam (n = 4). All panel members were medically qualified and most were psychiatrists. The panel comprised 20 males and 8 females. The ages of the panel members were: 13 aged 30–39 years, 13 aged 40–49 years, and 2 aged 50–59 years. Information was not collected on the clinical experience of the panel members.
Development and administration of the questionnaire
The method of developing the questionnaire has been described previously . Briefly, the content was based on a systematic search of websites, books, carer and consumer manuals, and journal articles for statements about how to help someone who may be experiencing a psychotic episode. These statements were grouped based on their common themes and used by a working group to generate questionnaire items specifying what actions a first aider should take. No judgements were made by the working group about the potential usefulness of the statements. Anything was included that fitted the definition of first aid, even if contradictory to other statements. The questionnaire was organized into sections of items on a common theme. These sections covered: recognizing and acknowledging that someone may be experiencing psychosis, encouraging the person to seek help, helping the person, interacting with the person, how to respond if the person becomes acutely psychotic, how to respond if the person becomes aggressive, and how to respond if the person denies they are unwell and/or refuses to get help.
The questionnaire was preceded by the following instructions: "Please complete the questionnaire by rating each statement according to how important you believe it is as a potential standard for Mental Health First Aid for psychosis. Please keep in mind that the standards will be used by the general public and as such, the statements need to be rated according to how important each one is as a way for someone, who does not necessarily have a medical or clinical background, to help a person who may have psychosis." Each statement was rated on the following scale: Essential, Important, Don't know/depends, Unimportant, Should not be included.
The questionnaire was administered as a web survey using the SurveyMaker application (surveymaker.com.au), with the option to complete it by email or paper mail if this was not possible. In Round 1 of the survey, panel members were asked to respond to the initial 219 items and were given an open-ended question at the end of each section asking for comments. If the comments included any new first aid statements, these were used to develop additional items for next Delphi round.
An item was accepted for inclusion in the guidelines if at least 80% of panel members rated it as "essential" or "important". If 70–79% gave this rating, the item was included for re-rating in the next round, with feedback to panel members about what percentage rated it highly in the previous round. This process continued for three rounds, which gave the opportunity for any new items to be rated and, if necessary, re-rated.
Once the list of consensus items was complete, these were used to write a piece of continuous text incorporating the first aid statements. This text was sent back to panel members for final endorsement and any suggestions for improvements. The final version constituted the guidelines.
Ethics approval was obtained from the University of Melbourne Human Research Ethics Committee (Project No. 0605537).