An exceedingly high variability with regard to the utilization of health care resources has been documented in many settings. Some of the variability is caused by practices based solely on physician style rather than evidence regarding efficacy of diagnostic or therapeutic interventions. Physicians are particularly poor at effective use of the medical laboratory. The problem has only been exacerbated by the continued proliferation of new diagnostic technologies. Computerized knowledge bases, which can support physicians requirements for timely access to medical information, have been proposed as a partial solution to this problem. The current proposal is a randomized, controlled trial of the effectiveness of the Quick Medical Reference System (QMR), a computerized medical decision support system designed to aid physicians in the process of medical diagnosis. The main purpose of this study is to determine the effect of the QMR program on the process and outcome of medical care provided to patients admitted to general medicine wards. Process and outcome of medical care provided to patients admitted to general medicine wards. Process and outcome measures will include accuracy of diagnosis, time until correct diagnosis, total length of hospital stay, diagnostic services utilized and total charges incurred. With the increasing proliferation of sophisticated and costly technologies, physicians especially those in training, have come to rely on information from expensive and often invasive testing, rather than to extract all the predictive potential of the history and physical examination. Also, many tests are ordered without thought given to the test's ability to help discriminate among diagnostic possibilities (i.e. a test with low predictive value is often obtained. The hypothesis we wish to test is that the clinical introduction of a system which suggest diagnostic hypotheses, as well as potentially discriminating history, physical, and simple laboratory items could result in more rapid and more efficient diagnosis and thereby lower the cost of care. A related sub-aim of the study, which can be accomplished with little additional data collection, is to determine the relationship of diagnostic uncertainty to the cost of diagnosis, and to the other measurements listed above. Uncertainty of diagnosis has been postulated as a heretofore unstudied variable affecting the higher cost of care rendered in teaching hospitals.
