The objective of this project is to implement a clinical prediction rule that calculates and displays the probability of true bacteremia for patients with positive blood cultures, and to determine if this information is helpful to physicians when they make treatment decisions. The clinical prediction rule uses a logistic regression model that stratifies positive blood cultures into 4 risk categories based on factors that are known at the time of first report. The prediction factors used to determine risk category were organism type, time until the culture first turned positive, and the presence of other cultures positive for the same organism. This rule has been implemented in the BWH's microbiology results review system so that the probability information is displayed whenever a positive blood culture result is viewed online. This study demonstrates that we can use electronic data to make calculations for a clinical prediction rule to help clinicians assess the probability that a given positive blood culture result represents true bacteremia. Aggregation of electronic data in ways that can help clinicians with decision-making is a key way that electronic records can improve care. Further evaluation is underway to determine the actual impact of this intervention on physicians' behavior and patient outcomes.