Clinical validity of alerts generated by the CareEngine Claims-Driven Decision Support Engine.

Clinical decision support based on administrative data is an important "real-world" challenge. At an academic medical center, we reviewed medical charts to assess the accuracy of 182 clinical alerts generated from administrative claims, pharmacy data, and lab results by a computerized clinical decision support (CDS) system. Overall accuracy was 82%. The primary factor affecting alert accuracy was availability of complete data. Claims-driven CDS may be sufficiently accurate to be a valuable adjunct to quality improvement.