ABSTRACT
This study was conducted as part of a two-state study of the validity of hospital discharge data for adverse event (AE) surveillance. Reliability of nurse identification of AEs was assessed through comparison with physician review. Nurses reviewed 6,296 medical records selected from two sampling frames. All were patients in 41 Utah acute care hospitals and surgery patients in 36 Missouri hospitals. A total of 1,035 records were independently reviewed by physicians using the same methods. Overall agreement between nurse and physician findings in Utah and Missouri, at the case level, was 78.5% (238/303) and 68.2% (499/732), respectively. Nurse review of medical records to identify AEs represents a cost-effective alternative to physician review for large-scale and ongoing patient safety improvements and surveillance purposes.