Towards the creation of a flexible classification scheme for voluntarily reported transfusion and laboratory safety events

To address this problem, we sought to create a “best-of-breed” patient safety classification for data contained in the Duke University Health System Safety Reporting System (SRS). Our approach was to implement the internationally recognized World Health Organization International Classification for Patient Safety Framework, supplemented with additional data points relevant to our organization. Data selection and integration into the hierarchical framework is discussed, as well as placement of the classification into the SRS. We evaluated the impact of the new SRS classification on system usage through comparisons of monthly average report rates and completion times before and after implementation. Monthly average inpatient transfusion reports decreased from 102.1？±？14.3 to 91.6？±？11.2, with the proportion of transfusion reports in our system remaining consistent before and after implementation. Monthly average transfusion report rates in the outpatient and homecare environments were not significantly different. Significant increases in clinical lab report rates were present across inpatient and outpatient environments, with the proportion of lab reports increasing after implementation. Report completion times increased modestly but not significantly from a practical standpoint.A common safety vocabulary can facilitate integration of information from disparate systems and processes to permit meaningful measurement and interpretation of data to improve safety within and across organizations. Formation of a “best-of-breed” classification for voluntary reporting necessitates an internal examination of localized data needs and workflow in order to design a product that enables comprehensive data capture. A team of clinical, safety, and information technology experts is necessary to integrate the data structures into the reporting system. We have found that a “best-of-breed” patient safety classification provides a solid, extensible model for adverse event analysis, healt