Mariken Stegmann

representatives (n = 18; 69%) completed the whole questionnaire or at least 70% of the  questions. The respondents’ characteristics met the inclusion requirements (Table 1).  Members of the panels were more mostly experienced clinicians and older patients.  Composition of the item list  The process for composing the original item list for panel review is shown in the  supplementary material (Supplement 1, Figure S1 b–c). In round one, the differences  between the median scores of GPs and specialists were never ≥2 points, so no weighting was  applied. In round two, the overall CVI for both letters was 71%, indicating that a third round  was not necessary. For those items for which no agreement was reached, the patient panel  was usually inclined to include the item, while the GPs and medical specialists were not.  Delphi procedure: Referral letters  For referral letter content, 72 items were included in the original list. The participants rated  12 items (17%) as highly relevant, 45 items (63%) as less relevant and 15 items (21%) as not  relevant (Supplement 2). Thus, the consensus list for referral letters consisted of 57 items  after the first round. All 12 highly relevant items from round one remained after round two,  but consensus did not reach the required level for any of the 45 less relevant items.  Therefore, only 12 items were included in the final consensus list for the referral letters  (Table 2). These items mainly concerned medical facts (e.g., past medical history,  resuscitation policy and medication details [names, doses and allergies]). Besides requiring  the reason for referral and the level of urgency, information was also required about the  presenting symptoms and the history of symptoms, as well as any aberrant findings on both  physical examination and investigation. Regarding psychosocial information, the only  retained item was the need for an interpreter when a language barrier was present.  However, there was no requirement for a provisional diagnosis or for treatment, contextual  or psychosocial information.  Delphi procedure: specialist letters  Of the original 108 items, 32 (30%) were rated as highly relevant, 51 (47%) as less relevant  and 25 (23%) as not relevant (Supplement 2). After the second round, all 32 highly relevant  items remained on the list and 3 less relevant items were included. Therefore, the final  consensus list for specialist letters contained 35 items (Table 2). Three new items were  proposed in round one, but each was only suggested once (Supplement 2).  In the final consensus list for specialist letters, the requirements were for the purpose of the  letter to be stated, for the corresponding medical facts to be included from the referral  letters (e.g., history, resuscitation policy, and medication details) and for details of any trial  enrolment. The requirement for information about the diagnostic process was limited to  details about presenting symptoms, physical examination (abnormal findings relevant to the  current problem) and investigation results (aberrant findings relevant to the current  problem). Concerning the diagnosis, it was agreed that the following six items should be  included: provisional diagnosis/diagnosis, whether the tumour is localised or metastasised, a  description of the size and local invasion of the primary tumour, whether there is spread to  regional lymph nodes, and whether there is distant metastasis.   7 85 Delphi consensus study of correspondence