Authors (including presenting author) :
Ng HL, Fujikawa T, KWOK WT, Wan S, HO YK, CHOW CY, CHAN WY, Lim K, CHANG TC, Siu CH, Wong HL
Affiliation :
Division of Cardiothoracic Surgery, Department of Surgery, Prince of Wales Hospital
Introduction :
The Dendrite Cardiac Surgery data collection system was established in 2005 at the Prince of Wales Hospital, New Territories East Cluster of Hong Kong. Since its introduction, yearly or biennial cardiac surgery reports were produced as part of the quality assurance and result monitoring mechanism of our cardiac surgical activities.
Objectives :
To develop a well-validated international risk-scoring system for the local population for long-term quality assurance and research purposes.
Methodology :
There are more than 6000 adult patient records accumulated in our database. To produce the biennial report on cardiac surgical outcomes for internal use, we collected data for all patients undergoing cardiac surgery for a 24-month consecutive period. Data was exported from a commercially available computerized system to conduct a more comprehensive data analysis. An ‘audit trial’ for data collection and input was established to ensure accuracy and validity on a weekly basis. Pre-operative data was collected to allow the calculation of the logistic EuroSCORE and EuroSCORE II, risk-stratification models validated in other patient populations. Following the calculation of the logistic EuroSCORE and EuroSCORE II, the predictive accuracy for the local population was assessed using the Receiver Operating Characteristic (ROC) curve (% true positive vs % false positives). Using the data in the system, risk-scored trend analysis was used to graphically analyze surgical performance. Standard CUSUM plots were constructed to record the cumulative sum of patient deaths over time. Variable Life-Adjusted Display (VLAD) graphs were generated to show the difference between the predicted and actual outcomes of all cardiac procedures performed. Funnel plots were common tools in comparing the morbidity rate (re-operation for bleeding, new post-operative stroke, and new post-operative haemofiltration) to international publication by the Society for Cardiothoracic Surgery (SCTS) in Great Britain & Ireland over time.
Result & Outcome :
Results: In 2019-20, areas under the curve were 0.808 (logistic EuroSCORE) and 0.839 (EuroSCORE II) respectively, indicating accurate discrimination of both scoring systems. The observed vs expected results for all isolated CABG surgeries (1.4% vs 3.8%), isolated valve surgeries (3.9% vs 10.1%) can be quite confidently predicted by the EuroSCORE. However, regarding major aortic surgeries observed and expected mortality (5.1% vs 22.3%), we believe a dedicated risk predicting system is needed for aortic procedure. Conclusion: With a sustainable increase in the data volume, we have implemented a mature and well-validated computerized database for patients undergoing cardiac surgery. Precise validation of a recognized risk-scoring system for the local population is further achieved by the weekly audit trial. We believe a dedicated scoring system for aortic procedure is needed for our locality.
