Predicting restenosis of drug-eluting stents placed in real-world clinical practice: derivation and validation of a risk model from the EVENT registry.

BACKGROUND: Prediction of restenosis after percutaneous coronary intervention (PCI) remains challenging, and existing risk assessment algorithms were developed before the widespread adoption of drug-eluting stents (DES). METHODS AND RESULTS: We used data from the EVENT registry to develop a risk model for predicting target lesion revascularization (TLR) in 8829 unselected patients undergoing DES implantation between 2004 and 2007. Using a split-sample validation technique, predictors of TLR at 1 year were identified from two thirds of the subjects (derivation cohort) using multiple logistic regression. Integer point values were created for each predictor, and the summed risk score (range, 0 to 10) was applied to the remaining sample (validation cohort). At 1 year, TLR occurred in 4.2% of patients, and after excluding stent thrombosis and early mechanical complications, the incidence of late TLR (more likely representing restenosis-related TLR) was 3.6%. Predictors of TLR were age <60, prior PCI, unprotected left main PCI, saphenous vein graft PCI, minimum stent diameter < or =2.5 mm, and total stent length > or =40 mm. Comparison of observed versus predicted rates of TLR according to risk score demonstrated good model fit in the validation set. There was more than a 3-fold difference in TLR rates between the lowest risk category (score=0; TLR rate, 2.2%) and the highest risk category (score > or =5; TLR rate, 7.5%). CONCLUSIONS: The overall incidence of TLR remains low among unselected patients receiving DES in routine clinical practice. A simple risk model incorporating 6 readily available clinical and angiographic variables helps identify individuals at extremely low (<2%) and modestly increased (>7%) risk of TLR after DES implantation.

Duke Scholars

Author Michael J Pencina Biostatistics & Bioinformatics, Division of Biostatistics