We propose to ask three related questions: 1) How effective are automated information systems for improving the use of ancillary tests; 2) What are some effective methods and clinical situations in which patient-specific information and probabilities can actively be given to providers to improve the quality of therapy in ways that do not disrupt the flow of patient care; and 3) How can an automated information system help an organization monitor quality of care? Because ancillary tests are often overused and because there is consensus for many regarding how they should be used, this proposal will focus primarily on how these questions relate to ancillary tests, on how providers respond to results for these tests, and in addition, how information systems can be used to identify adverse events. Our organization is in a particularly good position to ask these questions because we have a highly developed information system which is used and liked by providers, and because we have just implemented a provider order-entry system in which all orders are entered directly on- line, primarily by physicians. This system will be used to give providers information and feedback at the time orders are written, which should result in both improved quality and efficiency. First, we will use the information system to implement interventions for improving the use of ancillary tests, including both clinical laboratory tests and radiological procedures. We will serially perform four randomized controlled trials; the interventiOns will be: 1) display of charges for tests; 2) display of date, time and result when test last performed; 3) provision of reminders for redundant tests; and 4) structured ordering plus feedback when the indication for Performing the test is low yield. We will also evaluate the impact of actively providing patient-specific information and probabilities designed to improve the quality of therapy. We will perform two randomized controlled trials: in the first, we will use the computer to contact providers when a markedly abnormal result is obtained, and in the second, will give providers the probability that a positive blood culture represents a true positive. Finally, we will investigate two ways in which information systems may help organizations monitor quality. First, we will develop a computerized adverse event monitor, which will be used to define the rates for a number of sentinel adverse events. Second, we will develop novel ways for using the information system to provide concise, actionable feedback to management, using ancillary test use as a paradigm.