|
Back to 85th Annual Meeting
Back to Program Outline
3. Case Complexity Scores in Congenital Heart Surgery: A comparative validation study of the Aristotle Basic Complexity score and the Risk Adjusted Congenital Heart Surgery (RACHS) scoring system.
Osman O Al-radi, Christopher A Caldarone, Brian W Mccrindle, Gail Williams, William G Williams, Glen S Van Arsdell; Toronto, ON, Canada
Background:
The Aristotle Basic Complexity (ABC) Score and the Risk Adjusted Congenital Heart Surgery (RACHS) system were developed to compare outcomes of congenital cardiac surgical care among institutions and providers. Both ABC and RACHS are based on consensus methodology and have not been validated independently.
Methods:
Outcomes of all patients who underwent congenital heart surgery at a single institution from July 1982 to July 2004 (n=13675) form the Congenital Cardiovascular Surgery Database. Unique procedure codes were used to match the index operations (first operation during an admission) with ABC and RACHS scores, 13000 (96.2%) and 11954 (84.4%) procedures matched, respectively. Logistic regression models were generated with hospital mortality as the outcome and the respective score as the predictor variables. The ABC score was modeled both as a continuous variable (range 1.5 to 14.5) and a categorical variable (levels 1 to 4). Model goodness of fit, namely discrimination and calibration were assessed for each model with and without adjusting for time (surgical era). Discrimination was assessed by area under the Receiver Operator Characteristics (ROC) Curve (c-index), and calibration was assessed across the range of predicted and actual hospital mortality by bootstrapping.
Results:
ABC and RACHS are strongly associated with hospital mortality. When ABC is modeled as a continuous predictor the relationship between hospital mortality and the ABC score is not linear. After a score of 8, hospital mortality increases to a greater extent for a given change in the ABC score. The relationship is more linear in the current surgical era. Both ABC and RACHS scores had moderate discrimination, c-index 0.64 and 0.74, respectively. The calibration of models based on ABC as a continuous score, and based on RACHS improved when the model was adjusted for time. However, models based on ABC as a categorical level were inconsistent when adjusted for time.
Conclusions:
Both ABC and RACHS were associated with hospital mortality. The ability to predict hospital mortality with the two scoring systems was low to moderate. When used to compare centers or health care providers these scores should be calibrated to change in surgical era. Improvements in score discrimination would be beneficial.
Back to 85th Annual Meeting
Back to Program Outline
|
