The Delphi method
The Delphi method is an approach to transforming the opinions of individual experts into group consensus. It has been used in multiple fields including mental health research [20, 23,24,25]. In this study, the Delphi method was used to obtain consensus between mental health professionals and people with lived experience on appropriateness of statements to be included in the guidelines on helping a person at risk of suicide in Sri Lanka.
Study procedures included the following stages; (1) translation of English language questionnaire to Sinhala, the national language of Sri Lanka, (2) panel member identification and recruitment; (3) data collection over 2 rounds of survey; (4) data analysis and (5) guidelines development.
Translation of the English language questionnaire to Sinhala
The English language questionnaire (consisting of the items that were included in the final guidelines) was translated into Sinhala by a health professional. During this process, minor changes were made to some statements to make them more appropriate to the Sri Lankan health system and cultural context, e.g., emergency ambulance services which can be summoned by public are not available everywhere in Sri Lanka so this item was replaced with ‘calling Suwasariya (1990) ambulance service or organizing an alternative means to take the person to hospital’. Similarly ‘to see a GP’ was replaced with ‘to see the family doctor’, as this term is more common in Sri Lanka [26]. Because of the assumption that all the health workers can read and understand English and most of the consumers within the study areas can comprehend either Sinhala or English, Tamil translation was not done although that is the second most commonly used language in Sri Lanka.
The round 1 questionnaire comprised 168 statements categorized under 8 sections. Participants were asked to rate each item according to its importance for inclusion in the Sri Lankan guidelines on a five-point Likert scale with response options of, essential, important, depends/don’t know, unimportant and should not be included. The questionnaire also contained questions about socio-demographic characteristics, professional status and experience in mental health service provision (for health professionals).
Identification and recruitment of participants
Two groups of participants were recruited into the study: (1) Mental health professionals, (2) People with lived experience and caregivers (also referred to as consumers). Mental health professionals met eligibility criteria if they had been involved in providing mental health services for at least 2 years’ in either the state or private curative and preventive sectors. Inclusion criteria for consumers were as follows: (1) They had at least 1 year’s lived experience after an attempt of suicide; or (2) Or 1 years’ experience in caring for a person at risk of suicide.
Purposive and snowball sampling approaches were used to recruit professionals and people with lived experience. In order to maximize diversity of opinion (which is important for Delphi expert consensus studies) [27], we aimed to recruit participants from a wide variety of professional roles and from five different administrative districts across four provinces. This included tertiary and secondary level specialized Mental Health Units with in-patient care, primary level Mental Health Units providing out-patient and follow up care and public health institutes providing community- level care.
After identification of settings for recruitment, approval was obtained from relevant administrative authorities. For specialized mental health units (secondary and tertiary care), approval was obtained from the Director of the institute. One of the authors (AC) visited the units and directly approached participants, explained the purpose of the study and offered paper questionnaires. Participants were then free to decide whether or not to complete these. For primary level mental health units and public health institutes, the respective Regional Directors were approached for permission. AC visited these facilities during their monthly review meetings and directly approached participants.
In order to recruit people with lived experience, for each setting, a coordinator (typically a clinic nurse) was identified. This coordinator explained the purpose of the study and invited those eligible and interested in participating to attend a session in which the questionnaires were distributed and administered.
Data collection
In each setting, a short introduction was given to all participants, in which they were instructed to rate how important the statements were to be included in the guidelines. Health professionals were given the choice of whether to complete the survey in Sinhala or English, while consumers were given Sinhala questionnaires. Participants were also requested to add comments modifying existing statements or to suggest new items to be included. In recompense for their time, participants were given a gift voucher valued at Sri Lankan Rs 1500 for completing at least the Round 1 survey.
Data analysis
Statements were immediately included in the guidelines if they were endorsed by ≥80% of members in both panels as either essential or important. Statements were re-rated in the following round if they were rated as essential or important by 70–79% of either panels but were excluded if they were rated as essential or important by less than 70% of one panel.
Comments and suggestions from participants were refined, sorted and translated into English by one of the authors (AC) and then reviewed by authors, NR, AC and MF. New ideas were written into statements and included in the Round 2 questionnaire. The Round 2 questionnaire comprised 17 Items selected for re-rating based on the above-mentioned criteria and 14 newly generated items (a total of 31 items).
Participant socio-demographic characteristics, professional status and experience in mental health service provision were analyzed using descriptive statistics. Endorsement levels for each item were also calculated. The correlation between the endorsement rates of two panels of professionals and consumers was assessed using Spearman’s correlation coefficient. Analysis was done by SPSS Version 16.0 statistical software.
Guidelines development
Endorsed statements (i.e. those being rated as either essential or important by ≥80% of both panels) from both rounds were compiled. The Sinhala guidelines were developed by writing the list of endorsed statements into sections of connected text. Statements were amalgamated when possible. The language was changed in certain items to clarify meaning. The draft was then circulated to a panel of Sinhala speaking experts and non-health professionals for final review. Their inputs were included, and final guidelines were drafted.