February 1, 1996Circulation
Analysis and Comparison of Operator-Specific Outcomes in Interventional Cardiology
Key result
Operator-specific PCI models yield 0.85 ROC area for death but low complication rates limit discrimination.
- n=4,860
Why the study?
Do risk-adjustment models accurately discriminate operator-specific outcomes in percutaneous coronary revascularization?
Population
4,860 patients undergoing percutaneous coronary revascularization treated since January 1, 1990, from a…
Design
Cohort
Authors
Discussion
Loading...
Member takes
Implication
Operator scorecards for PCI warrant cautious interpretation; leaves open whether larger datasets or refined metrics can improve discrimination.
Key Points
- The aim was to develop methodologies for assessing operator-specific results in cardiovascular procedures.
- Utilized a multicenter database with 4860 patients from January 1, 1990, to assess provider outcomes.
- Developed models using multivariable logistic regression for evaluating outcomes such as death and myocardial infarction.
- Applied ROC analysis to evaluate predictive accuracy and identify operators' performance.
- Major complications occurred in 3.6% of patients.
- ROC areas were 0.85 for death, 0.77 for combined events, and 0.76 for procedural success, indicating moderate predictive accuracy.
- Provider outcome identification was inversely related to procedural volume (P = .05).
Study Design
Type
Observational (n=4,860)
Multicenter
Yes
Structured PICO
Do risk-adjustment models accurately discriminate operator-specific outcomes in percutaneous coronary revascularization?
P
Population
4,860 patients undergoing percutaneous coronary revascularization treated since January 1, 1990, from a multicenter database, involving 38 providers performing 25 to 523 procedures.
I
Intervention
Multivariable logistic regression models for risk adjustment and operator-specific outcome assessment ('scorecarding')
O
Outcome
Predictive accuracy (area under the ROC curve) and capacity to discriminate between superior and inferior provider outcomes for combinations of death, myocardial infarction, bypass surgery, and procedural successcomposite
Risk-adjustment models for assessing operator-specific outcomes in percutaneous coronary revascularization have modest predictive capacity and are limited by low event rates and low procedural volumes, suggesting scorecarding should be interpreted with caution.
Limitations
- Low incidence of major adverse events
- Subjectivity or susceptibility to manipulation of more frequently occurring adverse events
- Generally modest predictive capacity of the models
- Low volume of individual provider treatments
- low incidence of major adverse events
- subjectivity or susceptibility to manipulation of more frequently occurring adverse events
- generally modest predictive capacity of the models
- low volume of individual provider treatments
Cite This Study
Ellis et al. (1996) conducted an observational in percutaneous coronary revascularization (n=4,860). Operator-specific outcome modeling was evaluated on Predictive accuracy (ROC area) for death, myocardial infarction, bypass surgery, and procedural success. Operator-specific outcome models for percutaneous coronary revascularization yielded an ROC area of 0.85 for death, but discrimination was limited by a low 3.6% major complication rate.