Mariken Stegmann

Introduction  The incidence of cancer is rising in Europe [1,2]. Cancer care is complex in most instances,  requiring the involvement of multiple healthcare providers for diagnosis and treatment [3– 5]. Consequently, a patient may visit many different healthcare providers over a relatively  short and intense period.  In the Netherlands and other countries with a similar system, all inhabitants are registered  with a general practitioner (GP). This GP functions as gatekeeper and coordinator of  healthcare. In countries with such a system, patients often consult their GP for support [6,7].  To provide this care in an optimal manner, timely and efficient information exchange is  essential, both at time of diagnosis and during therapy [5,8,9]. As the electronic patient  records of GPs and medical specialists are separate, the most common way of  communication is the digital exchange of letters between GP (i.e., referral letters) and  specialists (i.e., consultation and treatment) [10]. These must be relevant and complete,  particularly with regards to the specific needs of the recipient party. Inadequate  correspondence by GPs may lead to unnecessary repetition of diagnostic tests, patient  anxiety and low patient satisfaction [4,11]. Timely and relevant information from specialists  can enable GPs to support their patient to make treatment decisions, give moral support  during treatment, and monitor and treat side‐effects and complications [10]. Despite all  parties agreeing that adequate correspondence is important, it is equally agreed that the  current situation could be improved [8,12].  We aimed to find consensus among GPs, oncology specialists (i.e., oncologists,  radiotherapists, urologists and surgeons) and patients regarding the content of referral and  specialist letters before and during the initial treatment (i.e. until start of treatment as  described in the original treatment plan) of cancer.  Methods  Study design  We used the Delphi procedure to reach consensus about the required content [13].  According to our institutional review board, no approval was needed because this study was  not subject to the Dutch Medical Research Involving Human Subjects Act. All participants  were informed about the aims of the study and agreed to participate.  Composition of the item list  The initial list of items was based on two sources: the code list of an earlier qualitative study  [14] and the results of a literature search. The literature search was performed in Medline  and Web of Science (Supplement 1, Figure S1a). The items derived from the qualitative study  and the literature search were discussed until consensus on every item was reached among  four members of the research group (MH, MS, JS and AB). To test the applicability, the item  list was presented to two GPs and two medical specialists in individual sessions.  Composition of the expert panels  The following three expert panels were formed by purposive sampling to optimise diversity:  a GP panel, an oncology specialist panel (i.e., oncologists, radiotherapists, urologists and  surgeons) and a patient panel (patients or close relatives). Potential participants were  7 83 Delphi consensus study of correspondence