After designing the Health Promotion Program, it was validated by 17 specialists using the Delphi technique from June 26 to December 6, 2018. Four key activities in the Delphi technique include identifying the problem, selecting experts, determining the group size, and performing Delphi rounds (20).
To identify the problem, Delphi questionnaire was designed by the research team based on the content of the Health Promotion program using the Grounded theory study, review of quantitative and qualitative texts, and a focus group discussion. The selection of specialists was performed by purposive and snowball sampling. The Delphi questionnaire was sent to them via e-mail or print of the questionnaire in person; and they were asked to review the written content and express their views on the strategies. The responses returned via e-mail, SMS, phone call, and face-to-face visits were reviewed and summarized. The summary report was re-sent to specialists who were allowed to change their own answers according to the report.
Participants were selected at five stages of preparation for selection, population determination by name, introduction of other specialists, ranking the specialists, and invitation of specialists. At the preparation stage for the selection, the disciplines and centers related to the research topic were identified, including the infertility center directors, gynecologists, reproductive health specialists, faculty members, center staff, and infertile couples. In fact, qualified people with a background in the issue were invited to Delphi study. At the population determination by name, the names of individuals who were identified at the first stage, were written. At the stage of introducing other specialists, we contacted the specialists listed at the previous stage; and a list of other specialists was identified by them. At the ranking stage, the identified specialists were ranked. At the invitation stage, higher level specialists were invited after ranking.
In Delphi technique, the number of participants is usually less than 50 and often between 15 and 20 (21). In the present study, 13 specialists (a gynecologist, three reproductive health specialists, three faculty members of midwifery, four masters of midwifery, and two bachelors of midwifery) as well as five participants with unsuccessful infertility treatment (a master of midwifery and her husband who was a traditional medicine specialist, a bachelor of midwifery and her husband, a master of law, and a nurse practitioner with a doctorate degree) were invited to participate in Delphi rounds. 17 out of 18 participants completed the study, and a bachelor of midwifery did not participate in the third round.
In the first round, the Delphi questionnaire was started based on strategies derived from the program and closed-ended and open-ended questions. Members were asked in the Delphi questionnaire to answer the closed-ended questions, review the texts and focus group in the qualitative section of study; and mention the gaps or other opportunities affecting the design of the health promotion program if there were not included in the questionnaire. After receiving the answers, the researchers converted the collected data into a fully organized questionnaire. The questionnaire was used as a research tool for the second round of data collection. In the round, the program was referred to specialists who were asked to examine the content in three fields, clarity (clarity of content and no ambiguity), validity (connection of details of the program with the subject) and usability (the feasibility of implementing the program in infertility centers of Iran) and express their opinions in open-ended responses to the program.
In the second round, each Delphi method participant received the second questionnaire, and was asked to rate the items summarized by the researchers based on the data provided in the first round according to the Likert scale (weak with a score of 1–3, medium with a score of 4–6, and high with a score of 7–9). On this basis, it might be necessary for Delphi method participants to use a ranking scale in order to create initial priorities among the items; hence, the issues of agreement and disagreement could be identified (22) and a space could be created to identify new ideas, correct, interpret, eliminate and explain their strengths and weaknesses. In some cases, participants were even asked to express their reasoning and prioritization of options. In this round, the panel members' motivation for higher participation increased because they received feedback from their answers and were eager to determine the quality of their colleagues' answers (23). During this round, members participating in the Delphi Method expressed their levels of agreement on the program strategies using the Likert scale.
In the third round, each Delphi member received a questionnaire including cases and rankings summarized by researchers in the previous round, and they were asked to reconsider their views and judgments or state their reasons for non-consensus (22). Here, new predictions were provided by ranking the comments; and statistical summaries of answers were prepared. From this round on, members reconsider their responses and those of others; and the process was repeated until all new ideas were presented and the strengths and weaknesses of all opinions were identified. In this round, statements of the previous round were presented along with a summary of other findings in the group. In this round, the group members could change their opinions according to opinions of other group members (23). It should be noted that number of Delphi repetitions in the degree of consensus were largely depended on the researchers and could be from 3 to 5 (22).
During this round, members of the Delphi method received a questionnaire, including cases and rankings summarized by researchers in the previous round, and were asked to reconsider their opinions and judgments, reconsider their answers and those of others, and explain their reasons for disagreement. It should be noted that the Delphi method continued for three rounds until reaching a consensus in the present study, and finally the consensus views were presented as a final plan. In results of studies, a different range of consensus level from 51–100% was reported (24). In the present study, the consensus level was considered to be at least 70%. The consensus level of each statement was obtained in a way that the maximum number of members, who responded to one of the Likert numbers, was divided by the total number of members participating in Delphi method; and the resulting number was multiplied by 100.