A theory-based approach to analysing conversation sequences.

AIMS To assess the quality of communication generally two procedures are used: one defines categories of utterances and counts their frequency, the other uses global observer ratings. We investigated whether a sequence analysis of utterances yields results which more precisely reflect the process of a conversation. METHODS We re-examined data from a randomised controlled intervention study in which residents' interviews with simulated patients were analysed with the Maastricht History and Advice Checklist (MAAS-R) and the Roter Interaction Analysis System (RIAS). Using the U-file of the RIAS we studied the effect of different types of physician questions (open, closed questions, facilitators, other physician actions) on the length of uninterrupted patients' speech and content of utterances. We investigated also whether reciprocity indices improve after a communication skills training, and whether they correlate with global scores form MAAS-R. RESULTS Patients respond to a closed question with a mean of 1.78 (+/- 1.49) utterances as compared to 2.75 (+/- 2.72) utterances after an open question. The likelihood of a concern was more than 10 fold higher after an open question compared to closed questions. Reciprocal sequences make up less than 2 percent of the conversation, Still, they correlate with global items form MAAS-R. The 'empathy index' improves after the training.