Abstract WP403: A New Risk Index for Predicting Outcomes Among Patients Undergoing Carotid Endarterectomy in Large Administrative Datasets.
Background: Administrative datasetsare used frequently without risk adjustment to determine regional and national level characteristics pertaining to performance of carotid endarterectomy (CEA). We developed and validated a new index to provide risk adjustment and to predict patient mortality or other outcomes in the hospital in patients undergoing CEA.
Methods: The primary end point was occurrence of stroke, cardiac complications, or death during the post procedural period of CEA derived from the Nationwide Inpatient Sample (NIS) which is representative of all admissions in the United States. Multivariate logistic regression was performed to identify the effect of clinical and demographic factors associated with the composite (stroke, death, or cardiac event) end point. Data from 2005-2006 (study period 1) was used to derive the risk index score (weights used to define risk adjustment were identified from the logistic regression model) while data from 2007-2009 (study period 2) was used for validation of the risk index.
Results: A total 120,633 patients with mean age in yrs [±SD] of 71.1 [±9.5] and 42.4% females underwent CEA during the study period. The composite endpoint during study period 1 was 3.02%; cardiac event 1.77%, stroke 0.99% and in-hopital mortality 0.47%. Predictors of the composite endpoint were (odds ratio [OR], P value) as follows: age≥70 years (1.15, 0.013), atrial fibrillation (3.18, <.0001), CHF (1.81, <.0001), cigarette smoking (1.64, <.0001), symptomatic status (1.87, <.001) and chronic renal failure (1.64, <.0001). The receiver operating curve (ROC) of the risk index was 80% [±SE 5%] when applied to the validation set. Patients with scores 2-3 had higher composite rates of composite endpoint (OR 2.5, 95% confidence interval [CI] 2.3- 2.7), scores 3-4 (OR 4.8, 95% C.I. 4.3- 5.4) and scores of ≥5 (OR 7.4, 95% C.I. 6.4- 8.5) as compared to scores ≤1.
Conclusion: A new risk index is developed to assist in risk adjustment in the logistic regression model and is meant to be used with large administrative data sets to provide appropriate adjustment for comparative studies.
- © 2012 by American Heart Association, Inc.