Abstract 162: The Impact of Data Quality on Outcomes From The National Cardiac Surgery Registry in Australia - Can Derivable Data Help?

Abstract

Introduction: The Australian and New Zealand Society of Cardiac and Thoracic Surgeons (ANZSCTS) Database Program collects pre-, peri-, and post-operative clinical information on patients undergoing cardiac surgery across Australia. It is a clinical quality registry aimed at improving patient outcome by reporting and benchmarking hospital and surgeon performance. Reports for the Database is dependent on the capturing of accurate and complete data by trained data managers at 28 participating Units. The ANZSCTS Database was audited to determine the adherence of Units to the current data definitions of post-operative New Renal Failure (NRF). Methods: Data between 1st January 2009 and 31st December 2013 (N=45322) from the ANZSCTS Database Program was analysed for reported NRF in patients undergoing cardiac surgery each Unit. To assess the accuracy of the reported data, NRF was also derived using pre- and post-operative renal status variables according to the ANZSCTS Database Data Definitions criteria for NRF. The reported and derived NRF data was then evaluated by procedure type and stratified by Units. Results: Overall, reported NRF was 5.1% (2308/45129) compared to derived NRF of 4.9% (2223/45322) across the 28 Units over the 5 year period. In isolated coronary artery bypass graft, reported NRF was 3.6% (853/24017) compared to derived NRF of 3.4% (824/24081). Unit-specific evaluation showed large variation in reported versus derived NRF across the Units. Out of the 28 Units, 10 Units had under-reported NRF and 18 Units had over-reported NRF post cardiac surgery (see figure). Conclusion: This study demonstrates that Units may not always adhere to the data definitions when collecting and entering data into the Database. As such, derivable data is a good indicator of potential data entry error within Units. In the case of NRF, the overall rates of NRF are similar (both derived and reported), however, large variations exists when NRF rates are evaluated at Unit level. This demonstrates that reporting overall rates of NRF in ANZSCTS Database routine activities may not be as useful as providing Unit-specific data as feedback. Therefore in future, reported NRF rates will be cross-checked against derived NRF, and both overall and Unit-specific rates of NRF should also be included in future Annual Reports.