Developing an Adverse Event Reporting System Using Administrative Data

    loading  Checking for direct PDF access through Ovid



Numerous studies have shown the rates of voluntary reporting of patient safety events to be disappointingly low. This article demonstrates that targeted use of administrative data can substantially improve identification of adverse events. The design and pilot test of a reporting system at the University of Michigan Health System (UMHS) is detailed. The system identifies otherwise unreported adverse events and provides information about the preventability of these events. When fully implemented, the reporting system will become an integral part of the UMHS system for improving patient safety.


The patient safety indicator (PSI)-case review system (CRS) is based on the Agency for Healthcare Research and Quality PSIs and uses a diagnosis timing variable to ensure that only inpatients with a condition that was acquired after hospital admission are identified and referred to clinicians for investigation.


During a 4-week pilot, 66 PSI cases for 56 patients were identified. Sixty-four cases (96%) were unrelated to adverse events recorded in the voluntary patient safety reporting system. Clinicians reviewed 50 cases, and 43 (86%) had a confirmed event. Nineteen (44%) were deemed potentially preventable and were associated with selected infections due to medical care, postoperative sepsis, postoperative pulmonary embolism or deep vein thrombosis, decubitus ulcer, and accidental puncture or laceration.


Although a brief experiment, the pilot test demonstrated that the PSI-CRS can enhance the voluntary patient safety reporting system and identify quality improvement opportunities. As an important side benefit, the system can also help UMHS prepare for federal initiatives to use administrative data to identify and reduce Medicare reimbursement for inpatients with hospital-acquired complications.

Related Topics

    loading  Loading Related Articles