Tracing the decision-making process of physicians with a Decision Process Matrix

BACKGROUND: Decision-making processes in a medical setting are complex, dynamic and under time pressure, often with serious consequences for a patient’s condition. OBJECTIVE: The principal aim of the present study was to trace and map the individual diagnostic process of real medical cases using a Decision Process Matrix [DPM]). METHODS: The naturalistic decision-making process of 11 residents and a total of 55 medical cases were recorded in an emergency department, and a DPM was drawn up according to a semi-structured technique following four steps: 1) observing and recording relevant information throughout the entire diagnostic process, 2) assessing options in terms of suspected diagnoses, 3) drawing up an initial version of the DPM, and 4) verifying the DPM, while adding the confidence ratings. RESULTS: The DPM comprised an average of 3.2 suspected diagnoses and 7.9 information units (cues). The following three-phase pattern could be observed: option generation, option verification, and final diagnosis determination. Residents strove for the highest possible level of confidence before making the final diagnoses (in two-thirds of the medical cases with a rating of practically certain) or excluding suspected diagnoses (with practically impossible in half of the cases). DISCUSSION: The following challenges have to be addressed in the future: real-time capturing of emerging suspected diagnoses in the memory of the physician, definition of meaningful information units, and a more contemporary measurement of confidence. CONCLUSIONS: DPM is a useful tool for tracing real and individual diagnostic processes. The methodological approach with DPM allows further investigations into the underlying cognitive diagnostic processes on a theoretical level and improvement of individual clinical reasoning skills in practice.