Determinants of success of coronary angioplasty in patients with a chronic total occlusion: a multiple logistic regression model to improve selection of patients.

OBJECTIVE To study the determinants of success of coronary angioplasty in patients with chronic total occlusions, and to formulate a multiple logistic regression model to improve selection of patients. DESIGN A retrospective analysis of clinical and angiographic data on a consecutive series of patients. PATIENTS 312 patients (mean age 55, range 31 to 79 years, 86% men) who underwent coronary angioplasty procedure for a chronic total occlusion between 1981 and 1992. RESULTS Procedural success was achieved in 191 lesions (61.2%). A major complication occurred in six patients (1.9%). Multiple stepwise logistic regression analysis identified the presence of bridging collaterals (p < 0.001), the absence of a tapered entry configuration (p < 0.001), estimated duration of occlusion of greater than three months (p = 0.001), and a vessel diameter of less than 3 mm (p = 0.003) as independent predictors of procedural failure. The logistic regression model was used to classify patients into groups of high, intermediate, and low probability of procedural success with cut off points of 70% and 30%. The predictive value for procedural success (probability > or = 70%) was 91% (95% confidence intervals (95% CI) 83% to 96%) and predictive value for procedural failure (probability < 30%) was 81% (95% CI 64% to 92%). CONCLUSIONS Percutaneous transluminal coronary angioplasty of chronic total occlusions is associated with a low risk of acute complication. Procedural success is influenced by easily identifiable clinical and angiographic features and the multiple regression model described may help to improve selection of patients.