30 experts GPs should be selected in order to have at least 20 in the last Delphi round. A minimum of 15 experts GPs was required in order to have at least 10 in the last Delphi round. The experts GPs were selected on the basis of having a German nationality, being fluent in English (understanding/speaking/writing), being involved in clinical practice, in research, and/or in teaching activities, and having the willingness to dedicate the time to this method of discussion. Information regarding age, gender, years of practice, practice type, setting type, teaching activities, research activities, number of English articles, number of native language articles and level of English was collected for all these experts. The research team wasn’t a part of the participants group.
From January 2017 to December 2017 the panel of experts were individually contacted by email to receive the original WAI-SR and its translation into German. Each participant was contacted separately to avoid contamination which is the basic methodology for Delphi procedure.
Participants ranked translation from 1 (absolutely no agreement) to 9 (fully agreement). The participant should explain each rank under 7 with a short commentary and a proposition for a new version. Consensus was defined as at least 70 % of the participants rating 7 or above the consensual items. If an item was consensual then it was validated and no further work was needed on it. If an item was not consensual all commentaries were collected and we proposed a new proposition taking into account the maximum of proposals. Then we designed a new form with only the non-consensual items and sent it again to the group. This process was repeated until a consensual translation was found for each proposition. If an expert did not answer, we sent a second email, then a third. After three recalls and no answer he was considered as lost.
Back translation and cultural check
The final phase involved the translation in German to English (backward blind translation). This was achieved by an official translator (native English speaker) and a non-official translator. They had no previous knowledge of the original English WAI-SR. Subsequently, the authors of the study compared the back-translated version with its original version for linguistic congruence and cultural relevancy. Then a final German translation of the WAI-SR patient version, client version and its scoring sheet was issued.
Backward translations and cultural check
Backward translations and the cultural check point out several discrepancies in terms of linguistic congruence and cultural relevancy. Every difference between the original WAI-SR and the two backward translations were discussed. All results of backward translations and cultural check are in annex 5.
Strengths and limitations of the study
The Delphi procedure for translation has its own strengths and weaknesses. Nevertheless, it is considered as an accurate consensus technique in health research (18)(19). In this study there was no information bias as all data were sent separately by mail to all participants. Anonymity among the panel of experts avoids dominance effect, group pressure and limits conflicts of interest. The study followed the recommendations of proper methodology for the Delphi technique according to the literature (20). The panel size was appropriate with 18 participants at the beginning of the study and 12 participants at the end of the second Delphi round which is enough to have significant results. Only 7 participants answered the third Delphi round, which is insufficient to validate question 8 from the WAI-SR patient version. Lots of commentaries were collected which signs the interest of the participants for the study.
There was no selected bias. The sampling data showed that most of the participants were working in a group practice (80%), were settled in an urban environment (78%) and had teaching (78%) and research activities (78%). The diversity of their gender was adapted because there were as many women as men. This study was a qualitative one. As a consequence, the sample did not need to be
! 31 representative of a group but heterogeneous, which is the case. Experts were selected for their knowledge: they all were German GPs with a good level of English (70%).
Before this study, a German version of the WAI-SR existed (21). Several points are different: the use of a validated consensus method and a cultural check to ensure the semantic and conceptual homogeneity and the use of a WAI-SR physician version with 10 items.
Table of contents :
MATERIALS AND METHODS
BACK TRANSLATION AND CULTURAL CHECK
CHARACTERISTICS OF THE RESEARCH TEAM
CHARACTERISTICS OF THE PANEL OF EXPERTS GPS
FIRST DELPHI ROUND
SECOND DELPHI ROUND
THIRD DELPHI ROUND
BACKWARD TRANSLATIONS AND CULTURAL CHECK
STRENGTHS AND LIMITATIONS OF THE STUDY
IMPLICATIONS FOR PRACTICE, MEDICAL TRAINING AND FURTHER RESEARCH