Healthcare-Associated Infection Decisions of Antibiotic-Resistant Organisms: A Data Quality Review

Background: Infection Prevention and Control (IPC) for Alberta Health Services and Covenant Health in the province of Alberta, Canada conducts surveillance for methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococcus (VRE) on all individuals admitted to acute-care and acute tertiary-care rehabilitation care facilities. Objective: The objective of this study was to determine the consistency and accuracy of infection decisions for MRSA and VRE. Methods: Surveillance cases of antibiotic-resistant organisms (AROs) collected using the provincial data entry surveillance platform between April 1, 2015, and March 31, 2017, across the province were reabstracted by infection control professionals and physicians using the NHSN infection definitions and compared to the original case severity decisions. Interrater agreement (Cohen’s ) and validity (sensitivity, specificity and predictive values) were calculated to compare the original and reabstracted infection decisions. Results: Collectively, 97% (87 of 90) of the IPC program staff and physicians who were initially invited re-abstracted 264 MRSA cases and 103 VRE cases within the review period. Provincially, 20% of the ARO cases reviewed (74 of 367) differed from the original infection decision. Among these 74 cases, 46 cases (34 MRSA and 12 VRE cases) changed from infection (original decision) to colonization (reabstracted decision) and 28 cases (21 MRSA and 7 VRE cases) changed from colonization to infection. The Cohen values for MRSA and VRE were 0.55 and 0.56, respectively, suggesting a moderate level of agreement for decisions made among IPC program staff. The sensitivity of the infection decision was higher with MRSA (86.5%) than for VRE (74.1%), meaning that there were more MRSA cases than VRE cases classified as infection in the original decision that remained infection following the review. Conclusions: Observed discordances on infection decisions were identified and may be attributed (1) to variations in the interpretation of the NHSN definitions, (2) to additional information that may have been available in the re-abstracted review compared to the original review, or (3) a difference in the information that was accessed to perform the original review compared to the reabstraction. This data-quality review provided an opportunity for IPC staff and physicians to become more familiar with infection definitions and such reviews will continue to be a regular process used to assess data quality within the IPC department.Funding: NoneDisclosures: None