METHODS
Twelve third-year students on their internal medicine clerkship at the University of California, San Francisco (UCSF)/San Francisco General Hospital and 14 teachers (8 residents and 6 attendings) were observed for 160 hours, including 73 oral presentations on rounds (42 by students and 31 by interns or postgraduate year 2 [PGY-2] residents ). Observation was by a trained rhetorician (LAL) who made rounds and took call with the patient care teams for part of two 8-week clerkships (October and November 1997, and January and February 1998). Nonparticipant observation was conducted following standard ethnographic technique,14 in which the observer dwells in the research community and, without engaging in the activities under study, records those activities and the relations between research subjects.
Observation was separated in time to allow detection of possible differences in presentation skills later in the clerkship year. The first group, a convenience sample consisting of 4 of the 8 students on the clerkship (2 students on each of 2 teams), was selected to allow in-depth observation of a small number of students and their teams (2 interns, 1 PGY-2 resident, and 1 faculty attending for each team). Students were observed during all activities of the clerkship for a 3-week period (approximately 100 hours). During this time, the mean number of observed presentations was 7.5 per student and 5 per house officer. Based on the data gathered from the first group, hypotheses were generated, and all 8 students on the clerkship during the second time period were observed (mean number of observed presentations was 1.5 per student) for a 2-week period during team work rounds, attending rounds, and/or presentation rounds with the clerkship director (approximately 60 hours); most oral presentations occurred in these settings. Saturation sampling (when further observations yield minimal or no new information15,16) was achieved through this process. Subjects were informed of our interest in “how students adjust to the clerkship”; however, in order to minimize observer effect, we did not disclose our specific interest in their communication skills until after the observation period.
Discourse-based interviews17 of 8 students and 10 teachers (5 residents and 5 attendings) were conducted and audiotaped. This sample included all the students on the clerkship during the second observation period and 5 of the 6 PGY-2 residents and 5 of the 6 team attendings during the same time period (those who agreed to be interviewed [all] and could be scheduled). Discourse-based interviews elicit tacit knowledge about language by having participants work with a discourse sample and explicitly justify content and organizational choices. Students were asked to arrange a written sample of patient material into oral presentation formats for different contexts and to justify and explain their choices. Teachers were given an already organized presentation sample and asked if they would present it differently in different contexts and to explain their choices. Teachers were also asked to interpret representative feedback statements selected from observational field notes. Different formats for the student and teacher interviews were chosen to reflect the preceptor relationship between students (creating the presentation) and teachers (critiquing the presentation). All students in the second group (n = 8) also completed a postclerkship survey. General survey questions inquired about the difficulties students had in composing and delivering case presentations, the “golden rules” of case presentation they had learned in their clerkships, and the advice they would offer to clerks beginning this rotation. Table 1 describes selected demographic characteristics of study subjects compared with the UCSF reference groups from which they were drawn. None of the teachers in the study had specific training in teaching oral presentation skills.
Table 1
Table 1
Characteristics of Study Subjects and Reference Groups, %
Data from field notes and transcribed interviews were qualitatively analyzed for emergent themes in order to uncover recurrent patterns of communication. Analysis followed the method of grounded theory technique15 in which textual data is organized into increasingly refined categories representing recurrent (“emergent”) themes. Final categories are checked with an expert insider (RJH) to ensure that they reflect the experienced reality of the discourse under study. Thematic findings from observations and interviews were triangulated using analyses of curricular documents, student surveys, and a review of the sociological, anthropological, and medical literature on medical discourse. Triangulation, a term from cartography, refers to the practice of collecting data from various sources in order to verify the accuracy of observational findings.
Although there were variations in presentation skills within our student and teacher groups (e.g., students demonstrated differing abilities, over time, to learn to adapt content to context), even after saturation sampling we did not find recognizable patterns to these intragroup differences. In addition, there was no discernible overlap between student and teacher presentation skills at any time in the study. Therefore, intragroup differences are not presented, and only intergroup comparisons between students and teachers are reported. Because no substantial differences were observed for students or teachers between the 2 time periods, the results were combined for analysis. Informed consent was obtained from all participants and the study was approved by the institutional review board of UCSF.