The purpose of the proposed project is to investigate the factors that might influence the diagnostic performance of physicians who use diagnostic decision support systems (DDSS). This study will extend the Principal Investigator's current research. A major finding of the current study was that the DDSS produced inappropriate diagnostic suggestion, as judged by an expert committee, even when the DDSS also included the correct diagnosis. The misleading suggestions could lead to adverse patient outcomes as well as increased cost of diagnostic work-up. Systematic research can indicate how physician users respond to the data from the DDSS. The present proposal is a plan for conducting such research.
The specific aims ar to (1) examine the relationships among the quality of information provided by a DDSS, the difficulty of a medical case, physicians, diagnostic performance using a DDSS,physician's clinical experience, physicians; experience with the DDSS, type of interaction the physician has with the DDSS, and physicians's perceptions of DDSS helpfulness; and, (2) to further evaluate the psychometric properties of the test cases used for testing the DDSS. The study design will be a nationwide stratified survey of physicians who will each attempt to diagnose eight medical cases using a DDSS. The sample will consist of 120 physicians from the U.S. who are current users of QMR. A total of 24 cases from the current study will be used. Each participant will be sent a set 8 cases, half of which are cases where the correct diagnosis in not in the QMR knowledge base. They will use QMR to assist them in arriving at a differential diagnosis. For each case the participants will electronically record their differential, their assessments of the relevance of the diagnoses produced by QMR, their assessments of case difficulty and helpfulness of QMR, as well as the specific nature of their interaction with QMR. Their differential diagnoses will be scored for accuracy and relevance using the scoring systems developed in the present study. Two clinical experts who reviewed the DDSS performance in the previous study will evaluate the participants' differential diagnoses. Multivariate analyses will be used to examine the relationship among case difficulty, relevance of the QMR suggestions, the participants; clinical experience/expertise, participants differential diagnoses, and the participants perceptions of case difficulty and helpfulness of QMR. This research will identify the factors which influence physicians' use of the data they obtain from a DDSS and can lay the groundwork for field trials of the use of DDSS.
