Impact of Monthly Data Quality Checks for Facilities on Surgical Site Infection (SSI) Data

BACKGROUND: Surgical site infections (SSI) following colon (COLO) and abdominal hysterectomy (HYST) procedures have been reportable to the Tennessee Department of Health (TDH) via the National Healthcare Safety Network (NHSN) since January 2012. Because resources are not sufficient to conduct onsite validation of SSI data in all Tennessee facilities, we developed quality checks to identify potential SSI reporting errors and provide each facility with a list of issues to investigate and correct. We sought to quantify the impact of these automated quality assurance measures on SSI data in Tennessee hospitals.

METHODS: Data quality analyses were developed using SAS 9.3 which check SSI event and procedure data for completeness, including timely reporting and complete SSI events and procedures. Checks also include potential data entry errors: incorrect infection detection timing, misclassification of outpatient/inpatient status of procedures, outlier procedure times, and outlier infection rates. Special focus was placed on identifying procedures with missing or outlier values for variables required for risk adjustment, because these procedures would be excluded from standardized infection ratio (SIR) calculations. Errors identified by these analyses are included in a set of HAI quality reports sent to facilities on a monthly basis. Support is provided to facilities that need assistance addressing errors, and follow-up is conducted for facilities with errors that are not corrected.

RESULTS: Of 96 facilities reporting COLO and HYST procedures to TDH between January 2012 and September 2013, 70 facilities (73%) had at least one error. Of the 29,808 COLO and HYST procedures performed during the same time period, 151 procedures (0.5%) were excluded from SIR calculations. Ninety-five procedures were excluded from the SIR because of missing or incomplete values for one or more variables required for risk adjustment. Fifty-six procedures were excluded because procedure time was missing, less than 5 minutes, or an outlier. Of the 70 facilities with errors, 38 (54%) had at least 1 procedure excluded from their SIR, and 4 facilities had 10 or more excluded procedures.

CONCLUSIONS: Performing automated data quality checks monthly and providing the results to facilities provides an opportunity to identify and correct errors that have a direct impact on facilities’ publicly reported SSI data. While the 151 procedures identified as being excluded from SIR calculations represent a small proportion of the procedures performed statewide, the potential impact on facility-level SIRs is significant. These quality checks can be modified and shared with other states to improve the quality of their facilities’ SSI data.