OBJECTIVES: To develop a reliable and valid computer coded measure to assess emotional expression from transcripts of physician-patient interactions. METHODS: Physician encounters with two standardized patients (SPs) were audiotaped. Fifty patients from each physician (n = 100 primary care physicians) completed surveys that assessed patients' perceptions of their relationships with physicians. Audio-recordings of 193 patient-physician encounters were transcribed and computer-coded to derive a percent emotion words, and research assistants completed the Measure of Patient-Centered Communication (MPCC). RESULTS: After adjustment for potential confounders, regression analyses revealed physicians' use of emotion words and the MPCC contribute independently to patients' and SPs' perceptions of their relationship with physicians. CONCLUSIONS: The computerized coding of emotion words shows promise as a reliable, valid, and simple method to code transcript data of physician-patient interactions. This method may be expanded to examine other aspects of physician language and does not require coder training.
OBJECTIVES: To develop a reliable and valid computer coded measure to assess emotional expression from transcripts of physician-patient interactions. METHODS: Physician encounters with two standardized patients (SPs) were audiotaped. Fifty patients from each physician (n = 100 primary care physicians) completed surveys that assessed patients' perceptions of their relationships with physicians. Audio-recordings of 193 patient-physician encounters were transcribed and computer-coded to derive a percent emotion words, and research assistants completed the Measure of Patient-Centered Communication (MPCC). RESULTS: After adjustment for potential confounders, regression analyses revealed physicians' use of emotion words and the MPCC contribute independently to patients' and SPs' perceptions of their relationship with physicians. CONCLUSIONS: The computerized coding of emotion words shows promise as a reliable, valid, and simple method to code transcript data of physician-patient interactions. This method may be expanded to examine other aspects of physician language and does not require coder training.