Prior research has used a variety of qualitative and quantitative approaches for studying handoff communication. Due to the dynamic and interactive nature of handoffs, characterizing the structure and content of these conversations is challenging. In this paper, we use a graph-based approach to characterize handoff communication as a conversation network. Conversation networks were used to compare the structural properties of resident-resident and nurse-nurse handoff communication. Resident (n = 149) and nurse (n = 126) handoff conversations from general medicine units were coded using a previously validated clinical content framework. The coded conversations were then translated into separate resident and nurse conversation networks, and were compared using 11 network measures. Transition probabilities were used to identify commonly repeating sub-networks within resident and nurse conversations. There were significant differences between resident and nurse conversation networks in 10 of the 11 network measures. There were also significant differences in the structure of conversations: compared to resident conversations, nurse conversations were focused on fewer clinical content categories and had more branching and switching between clinical content categories; however, there were clinically-relevant organic relationships in the order of presentation of clinical content among both resident and nurse handoff conversations. We discuss the potential for using graph-based approach as an alternative method for characterizing interactive conversations and also suggest future directions for using network-based approaches for analyzing handoff conversations.

Original languageEnglish
Article number103178
JournalJournal of Biomedical Informatics
StatePublished - Jun 2019


  • Conversation networks
  • General medicine
  • Graph analysis
  • Handoffs
  • Nurse communication
  • Resident communication


Dive into the research topics of 'A graph-based approach for characterizing resident and nurse handoff conversations'. Together they form a unique fingerprint.

Cite this