The exact meaning of "communication skills" remains unclear, and detailed information on the content of communication skills education is often lacking . Cegala and Lenzmeier Broz  reviewed 26 studies published since 1990 and found little consistency in definitions of communication skills in these studies. These authors note that, in many cases, the instruments used to assess communication effects miss the target of the intervention. We attempt to address this issue in the present study by carefully examining the literature on the assessment of communication skills to identify an appropriate measure for our training issues. To address the inconsistency in definitions, we use the Kalamazoo Consensus Statement  as a framework for our intervention. Published in 2001 by a group of representatives from major medical education and professional organisations, this statement provides a list of essential elements in patient-physician communication: (1) build the doctor-patient relationship; (2) open the discussion; (3) gather information; (4) understand the patient's perspective; (5) share information; (6) reach agreement on problems and plans; and (7) provide closure. We rely on the Kalamazoo statement because an equivalent for German-speaking countries had not been published at the time our intervention was planned.
Design and participants
At the University Medical Centre Hamburg-Eppendorf, an interdisciplinary team of faculty members is working on an overall reform of the curriculum based on the Bologna process, the implementation of a university education system consisting of bachelor's and master's degrees that are comparable throughout Europe. This reform will also include the revision of previous methods of teaching communication. In the course of this process, we developed and evaluated existing communication training to gain knowledge about its effective components as well as students' needs and preferences.
The communication-training module "Basics and Practise in Communication Skills" was embedded in a newly developed supplementary qualification for students in psychosocial-medicine at the University Medical Centre Hamburg-Eppendorf in Germany. We offered the newly developed training to a group of students who agreed to take part in addition to their regular studies (psychosocial-medicine students = PMS). The concept for our communication module was designed and field-tested in 2008 and was developed further in the following year.
The selection of course participants for the pilot project in 2008 took place by means of a lottery. Following an introductory event, approximately 150 interested applicants (from 400 students) submitted their names on lists that were made available in the lecture hall, from which 26 participants were selected. Over the course of the supplementary qualification training, one male participant discontinued involvement due to schedule conflicts. In the end, 25 participants participated in the course to its completion, of which 18 (72%) were women and seven (28%) were men. This gender distribution was not representative of the 150 students who volunteered (16% male), but it was nearly representative of the entire cohort (33% male). The average age was 23 years (SD 3 years). In 2009, students were asked to complete a written application, including a short CV and a demonstration of their motivation. From 26 interested students, 20 were selected to participate in the course in 2009, of which 14 (70%) were women and six (30%) were men (full cohort: 39% men). The participants' average age was 24 years (SD 3 years).
All of the students enrolled in the training were in the first year of their degree at the time of participation. The training took place over 19.5 h in 2008 and 33 h in 2009, which progressed over the same time frame as the semester.
To test the effects of the newly developed training against the established, standard course in patient-doctor communication, a comparison group was established. This group included second-year medical students who attended a standard communication course (standard curriculum students = SCS) scheduled in the second year (2008: N = 38, of which 25 (66%) were women and 13 (32%) were men; 2009: N = 13, of which 10 (77%) were women and three (13%) were men).
All students provided informed consent. The study was conducted in full accordance with the 1975 Declaration of Helsinki and the revised version of 1983 and in full accordance with national ethical guidelines.
Both groups completed self-evaluation questionnaires on their communicative skills at the beginning and end of the course. In 2009, a role-play (approx. five minutes) with simulated patients was videotaped for each student at the beginning (t0) and end (t1 for SCS, t2 for PMS) of the course and was subsequently rated to provide a more objective measure in addition to the self-evaluation questionnaires. The PMS group had another videotaped consultation in the middle of the course (t1). At this measurement point, the number of lessons taught was the same for both groups (see Figure 1).
Content of the module
Students were given the opportunity to develop their basic communication skills in peer role-playing scenarios and role-playing sessions with simulated patients. The use of simulated patients or actors has proven to be a particularly effective method for training communication skills cf. [30, 31]. Parallel to participation in these exercises, the students developed guidelines on physician communication, which covered the core elements in patient-physician communication according to the Kalamazoo Consensus Statement . These guidelines were then used as a basis for further teaching to provide the students with a manageable tool focused on the primary issues of good patient-physician communication. The guidelines included short descriptions of skills relevant to each of the following domains: (1) building the doctor-patient relationship; (2) opening the discussion; (3) handling emotion; (4) exploring details; (5) reaching agreement on further procedures; and (6) summing up the consultation.
During a subsequent lesson, the students led conversations to determine medical histories with simulated patients. An introduction to psycho-oncology provided the students with a glimpse into the emotional reactions of patients with severe physical illnesses through the use of selected case studies, allowing students to broach the issue of dealing with these reactions. Another lesson addressed the conveyance of the pros and cons of illness-related decisions (risk communication). Here too, case studies and video recordings of doctor-patient communication served as the basis for the examination and discussion of the exchange of information concerning prognoses or probabilities of health-related issues. The presentation of case studies through cooperation with a clinic for the treatment of patients with mental illnesses was received particularly well. Three to four students had the opportunity to take a patient's medical history in the presence of a physician. A further conversation with a patient from the oncological outpatient clinic occurred in a subsequent session. These exercises provided the students with the opportunity to utilise and expand upon the conversation techniques acquired during the course. Moreover, the assessment of a psychopathological diagnostic report was practised at the conclusion of the module using the previously discussed case studies.
The content of the module was similar for most of the lessons in 2009, but, in contrast to the limited opportunity in 2008 to conduct a medical history with a patient, all participants had the opportunity to conduct an interview with a patient from a psychosomatic clinic. In preparation for these interviews, students were taught basics of psychosomatic medicine, including common disorders and their prevalence. Furthermore, students practised writing an elaborate case history report for psychosomatic patients.
The course offered to the comparison group was the same in 2008 and 2009. This course comprised four sessions (12 h in total) of communication training, including exercises, peer role play, structured feed-back and encounters with standardised patients (see Table 1).
Two methods were used to measure improvement in communication skills. First, a self-rated questionnaire was used for evaluation. This questionnaire included two sections: first, a general section on socio-demographic and degree-related information (age, gender, semester and previous professional experience); second, a self-assessment of students' communicative skills and an evaluation of the course lessons. The evaluation was performed on a 6-point Likert scale (1 = very good, 2 = good, 3 = satisfactory, 4 = uncertain, 5 = unsatisfactory to 6 = insufficient). This self-rating was administered to both cohorts: the pilot phase in 2008 and the further evaluation in 2009.
In addition to the questionnaire described above, videotapes were evaluated by trained observers (2 advanced psychology students) using the Maastricht History-Taking and Advice Checklist (MAAS-Global)  for the 2009 cohort. The MAAS-Global is a well-established, objective measure for assessing communication skills [33, 34], which has been validated in several studies [35, 36]. For our study, we chose a modified version of the scale for use with medical students.
The original instrument consisted of 13 items concerning the course of consultation (e.g., introduction, physical examination or diagnosis) and specific communication skills (exploration, emotion, information giving, summarisations, structuring and empathy) . All items concerning the course of consultation were omitted because their content was not relevant to the evaluation of the communication skills course; the primary target of the training was not to conduct an interview in the correct order but to elicit specific communication skills. The items were rated on a 6-point Likert scale (1 = very good, 2 = good, 3 = satisfactory, 4 = uncertain, 5 = unsatisfactory to 6 = insufficient). In addition to these specific items, a subjective overall rating was collected. Raters were blind regarding group membership and the recording time of the videotapes. After completing the individual ratings, consensus ratings for each subject were performed. These ratings were used for all further analyses.
Regarding the reliability of the instrument used, we calculated the intra-class coefficient (ICC) as an adjusted measure of agreement. Fleiss and Cohen  showed that the ICC is equivalent to a weighted kappa for measures of reliability, and Landis and Koch  provided "rules of thumb" for the interpretation of kappa coefficients. According to these rules, kappa values between .21 and .40 are "fair," those between .41 and .60 are "moderate," those between .61 and .80 are "substantial," and those between .81 and 1.00 are "almost perfect." An examination of the effectiveness of the communication module was performed for both groups (PMS and SCS) using repeated multivariate variance analyses (ANOVA). A partial eta-squared measure was used for the effect size. The interpretation of the effect size is based on the recommendations by Cohen: partial eta-squared < .0099 (small effect), partial eta-squared > .01 and < .0588 (medium effect), and partial eta-squared > .1379 (large effect) .
The calculations were performed using SPSS 15 .