Course Level: Intermediate
Format Requirements: The course will consist of interactive lectures, with case examples and thinking problems to illustrate issues in diagnostic performance, problem solving exercises to learn how to apply normative rules in a diagnostic task, and discussions regarding the psychological processes which affect the quality of diagnosis and regarding the pros and cons of methods used to measure clinicians’ diagnostic performance in empirical research. Basic information will be presented as power point slides (including handouts of the relevant formulas for normative diagnosis). Case examples will illustrate psychological processes which sometimes get clinicians into trouble. There will be problem solving exercises to provide experience with the application of the normative diagnostic principles, with worked solutions. The class will discuss issues of appropriate methodology, of psychological processes which may affect diagnostic performance, and other questions relevant to the psychology of medical diagnosis that may come up.
Background: The course addresses the psychology of clinicians’ diagnostic performance, in terms of both accurate categorization of the cause of the patient’s complaint, and adjustment of the probability the patient has a diagnosis given new information. Understanding good diagnostic performance requires clarity on the nature of the diagnostic task and the best way to use information diagnostically, and a vocabulary for describing how clinicians do these tasks and for measuring their shortcomings. Attendees are shown the normative rules (for categorizing based on symptom diagnosticity, for updating probabilities with Bayes’ theorem, for deriving test thresholds or action thresholds based on utility of missing diagnoses or unnecessarily treating for them) and learn how to apply those rules in diagnosis. Moreover, they learn to judge the suitability of research methods for characterizing what clinicians do, and how clinicians fall short of the best diagnostic reasoning, in situations where those rules apply.
Description and Objectives: Diagnosis is an uncertain process based on imperfect information. Accurate diagnosis has potentially far-reaching consequences for the well-being of patients and the costs of health care.
In this course the psychological processes of clinicians’ diagnostic performance will be described, including both categorization of the cause of the patient’s complaint, and assessment of the probability the patient has a particular diagnosis. These descriptions will be complemented with a normative perspective on how these tasks should be performed, which includes optimal procedures such as rules for categorizing based on symptom diagnosticity, for updating probabilities with Bayes’ theorem, and for setting probability thresholds to trigger actions based on balancing the utilities of the errors of “miss” and “false alarm”.
The course provides an overview of available empirical evidence describing how clinicians perform these diagnostic tasks and what kind of shortcomings they show in these tasks. Attention to the shortcomings in clinicians’ diagnostic performance, compared with the optimal, can alert clinicians to common pitfalls in diagnosis; many errors can be attributed to heuristic reasoning strategies that most of us use without awareness. Focussing on shortcomings also helps us to understand how the clinician’s mind works and to come up with strategies to improve diagnosis.
The key objectives of this course are to:
- (A) clarify the nature of the diagnostic task and of diagnostic performance in terms of both processes and outcomes;
- (B) clarify normative rules regarding the best way to use information diagnostically in situations where those rules apply, and provide experience with their application;
- (C) provide a vocabulary for describing and measuring how clinicians do these tasks and an overview of available methodology for measuring their shortcomings;
- (D) provide an overview of psychological and task factors that can affect physicians’ diagnostic performance.