Abstract P350: Predicting the Restenosis Benefit of Drug-Eluting vs. Bare Metal Stenting Prior to Angiography: A Tool for Patients and Clinicians
| Autor: | Robert W Yeh, Philip G Jones, Robert E Wolf, Sharon-Lise T Normand, David J Cohen, Kalon K Ho, Donald E Cutlip, Aaron D Kugelmass, John A Spertus |
|---|---|
| Rok vydání: | 2011 |
| Předmět: | |
| Zdroj: | Circulation: Cardiovascular Quality and Outcomes. 4 |
| ISSN: | 1941-7705 1941-7713 |
| DOI: | 10.1161/circoutcomes.4.suppl_1.ap350 |
| Popis: | Background: While drug-eluting stents (DES) are associated with lower restenosis rates compared with bare metal stents (BMS), they require prolonged thienopyridine use and may be associated with higher rates of stent thrombosis. Recent work has demonstrated that the absolute risk reductions of DES (i.e. the restenosis benefits) vary by clinical and angiographic features. However, to engage patients in the decision-making surrounding stent type, the benefits of alternative stents must be shared prior to angiography so that they can participate in a non-sedated state. To address this challenge, we created multivariable models based on clinical characteristics available prior to PCI to predict restenosis one-year after DES or BMS. Methods: Patients undergoing PCI in Massachusetts between 4/03 and 9/06 were identified and 1-year target vessel revascularization (TVR) outcomes determined. Parsimonious multivariable logistic regression models were constructed, using pre-angiographic data from NCDR versions 2 and 3, to identify clinical factors associated with TVR. The final models for DES and BMS consisted of 12 and 10 sociodemographic and clinical variables, respectively, chosen by backwards elimination while maintaining 95% of the predictive ability of the full models. We simulated random combinations of risk factors to determine the range of predicted restenosis benefit for DES vs. BMS. Results: A total of 19577 DES and 5255 BMS admissions were included in the development cohort and 9809 DES and 2597 BMS admissions were used in a validation cohort. The c-statistics for the models for predicting TVR after BMS and DES were 0.60 and 0.61 in the derivation cohorts, and 0.615 and 0.62 in the validation cohorts. Good calibration was observed in all models. Depending on the simulated clinical profile, the relative risk of TVR after BMS vs. DES ranged from < 1.0 to > 10. Conclusions: We generated multivariable models to predict TVR with DES or BMS based on clinical variables available prior to PCI that are comparable to previously published models that include procedural variables. The predicted relative risk of TVR for DES vs. BMS varies widely based on patients' clinical characteristics. These models may be useful in engaging patients in stent selection prior to PCI. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|