Real-Time Clinical Decision Making

decision making, realtime, anesthesia, diagnosis, monitoring, control, planning, attention, data overload, cogniti ve overload Objectives: To study the intraoperative behavior of anesthetists in order to characterize clinical decision making processes under realtime conditions. Design: Non-participatory observation and retrospective subject interview. Clinical behavior was captured on video, and annotated by voice commentary provided by shadow anesthetist not participating in patient care. Subjects were interviewed pre and post operatively, and approximately one week later, when selected events were reviewed. Settings: Cardiac operating theaters at 2 UK teaching hospitals. Su bjects: 12 anesthetists, of which 7 were consultants and 5 were senior registrars in training. Results: Analysis of the data resulted in the development of a behavioral model which yielded several key observations: Anesthetic decision making is primarily focused on physiological control decisions rather than diagnosis. Most clinical activity followed structured plans. When physiological systems were not directly measurable, control strategies either chose an indirect but observable parameter as a substitute, or attempted to estimate the unobserved parameter from multiple pieces of evidence. Cognitive loading may swamp attentional resource and result in perceptual errors while mcnitoring, and may be caused by plan reformulation or multiparameter estimation. Data overloading was not observed in the study. Information gathering is structured around anesthetic tasks rather than physiological systems. Conclusions: Clinical decision making is normally described in terms of the diagnostic process, but in real-time domains like anesthesia decisions seem to be control oriented and plan based.