Abstract 17832: Model for Predicting Amputation and Death in Patients With Critical Limb Ischemia Following Endovascular Treatment
Introduction: Patients with critical limb ischemia (CLI) are at high risk for amputation and death. However, there is sparse data on predictors of amputation and death among CLI patients who have undergone endovascular therapy. We developed a multivariable prediction model for amputation and death at 3 years in CLI patients receiving endovascular treatment.
Methods: This investigation is a sub-study from a clinical registry of consecutive patients undergoing lower extremity revascularization procedures in 2 integrated health care systems from January 1, 2005 - December 31, 2011. Detailed clinical and procedural data were extracted from chart review by vascular specialists. Inclusion criteria for this analysis were: 1) documented CLI, and 2) receipt of endovascular treatment. A prediction model for amputation and death at 3 years was developed using elastic net regularized regression with a Cox proportional hazards model, and the model’s performance was assessed.
Results: Of the 1858 patients in the parent clinical registry, 291 (15.7%) met study inclusion criteria. Fifty percent were female, 53.3% had diabetes, 12.7% had a glomerular filtration rate (GFR) <30 cc/minute, mean hematocrit was 38.5 (±5.4), 6.5% had a prior amputation, 57.7% had active tissue loss, and 9.3% had 3-vessel tibioperoneal disease. Amputation and death rates were 20.3% and 15.5%, respectively at 3 years. Of 20 pre-specified, clinically plausible variables, the following covariates comprised the model: age, diabetes, GFR, hematocrit, prior amputation, tissue loss, and 3-vessel tibioperoneal disease.
The model had an optimism-adjusted c-statistic for amputation and death at 3 years of 0.71, and performed well with regard to calibration (0.838 shrinkage of calibration slope from the original estimated slope based on 200 bootstrap resamples).
Conclusions: Age, diabetes, GFR, hematocrit level, previous amputation, tissue loss, and degree of tibioperoneal disease are predictors of amputation and death in this study population. The derived prediction model for this outcome performs well and should be clinically vetted to improve outcome prediction in CLI patients.
Author Disclosures: R.K. Rogers: Other; Modest; Adjudication Committee member for VOYAGER trial funded by Bayer, Steering Committee Member for trial funded by Astra Zeneca, Steering Committee Member for observation study funded by CSI. K. Josey: None. D. Brostow: None. D. Magid: None. A.E. Barón: None. T.F. Rehring: None. O. Jazaeri: None. R. Gupta: None. S. Jones: Research Grant; Significant; Agency for Healthcare Research and Quality, American Heart Association, AstraZeneca, Bristol-Myers Squibb, Daiichi Sankyo, Patient-Centered Outcomes Research Institute. M. Patel: Research Grant; Significant; AstraZeneca, CSL, HeartFlow, Janssen Research & Development, Maquet, Medtronic, NHLBI. Consultant/Advisory Board; Modest; AstraZeneca, Bayer, CSL, Genzyme, Janssen Research & Development, Medtronic, Merck & Co. A.S. Go: Research Grant; Significant; CSL Behring. R.W. Chang: None. B. Hill: None. M. Ho: None. T. Tsai: None.
- © 2016 by American Heart Association, Inc.