Effects of Decision Support Tools on Cardiac Telephone Consultation Process

Abstract

The Nursing Coordinators (NCs) at the University of Ottawa Heart Institute (UOHI) fields phone calls from patients who have been discharged and are undergoing home care procedures at a daily basis. The project described in this thesis aims to provide tools for the Personal Digital Assistant (PDA) that the NCs can use during the phone calls. The Cognitive Work Analysis (CWA) and Ecological Interface Design (EID) approach are used to identify the information requirements to design the system. Major challenges of the telephone consultation process that are additionally identified by literature review and interviewing the NCs included visibility of patients, individual differences, and lack of standardized procedures. A combination of decision trees and visualization techniques is proposed to aid the process. Implementation of decision trees would help unload mental workload especially accesses to "knowledge in the head" as well as facilitate expert knowledge transfer to less experienced nurses. Visualization tools display integration of multiple-cues from patients in an abstract nature and can be accessed by users at any point of decision process. <br /><br /> Preliminary experiment with static images showed that visualization tools helped the decision makers more when the judgement tasks were more complex. The effects of different types of decision support on the cardiac nurses in simulated telephone consultation processes were examined. The system improved the performance of the decision makers and induced different types of strategic behaviours: a standardized checklist, OLDCAR, induced more through assessment, the decision algorithms induced efficient and more detailed recommendation, and the semantic network symptom map induced information gathering more relevant to diagnosis. <br /><br /> The research also explored methodologies to examine multi-layered decision process, where many decision makers with varying expertise are involved in modeling the strategic behaviours. This type of process can be applicable when the primary decision makers do not monitor the work domain, but can be alerted when something goes wrong.