PURPOSE: Models have been developed to predict the likely outcomes of endovascular aneurysm repair (EVAR) for patients, based on a longitudinal Australian audit. METHODOLOGY: Mid-term progress of 961 Australian patients who underwent EVAR has been collected and used to develop predictive models for 17 outcomes. Stepwise forward logistic regressions determined the significant preoperative patient variables to be included in each outcome model. An interactive program was subsequently developed to allow surgeons to review the predicted success rates for patients about to undergo the procedure. Each model was assessed using a global goodness of fit test and was internally validated using bootstrapping. RESULTS: Eight pre-operative variables were included in the interactive model for 17 outcomes. The eight variables used were aneurysm size, age, ASA, gender, creatinine, aortic neck angle, infrarenal neck diameter and infrarenal neck length. The outcomes predicted included perioperative mortality, perioperative morbidity, mid-term survival and reintervention rates. All outcome models achieved reasonable goodness of fit, with the exception of the model for conversion to open repair (p=0.04). With respect to validation, survival, aneurysm related deaths, migrations, ruptures and conversions to open repair performed best in terms of predictive discrimination. Models for survival, migrations and conversions to open repairs performed best in terms of bias corrected R-squared index. The models with the smallest calibration error were 3 and 5 year survival, early deaths and mid-term type I endoleaks. CONCLUSIONS: An interactive model is available, which can assist vascular surgeons to evaluate the expected outcomes for a particular patient undergoing EVAR. The validated model is useful for counselling and pre-operative decision making.
PURPOSE: Models have been developed to predict the likely outcomes of endovascular aneurysm repair (EVAR) for patients, based on a longitudinal Australian audit. METHODOLOGY: Mid-term progress of 961 Australian patients who underwent EVAR has been collected and used to develop predictive models for 17 outcomes. Stepwise forward logistic regressions determined the significant preoperative patient variables to be included in each outcome model. An interactive program was subsequently developed to allow surgeons to review the predicted success rates for patients about to undergo the procedure. Each model was assessed using a global goodness of fit test and was internally validated using bootstrapping. RESULTS: Eight pre-operative variables were included in the interactive model for 17 outcomes. The eight variables used were aneurysm size, age, ASA, gender, creatinine, aortic neck angle, infrarenal neck diameter and infrarenal neck length. The outcomes predicted included perioperative mortality, perioperative morbidity, mid-term survival and reintervention rates. All outcome models achieved reasonable goodness of fit, with the exception of the model for conversion to open repair (p=0.04). With respect to validation, survival, aneurysm related deaths, migrations, ruptures and conversions to open repair performed best in terms of predictive discrimination. Models for survival, migrations and conversions to open repairs performed best in terms of bias corrected R-squared index. The models with the smallest calibration error were 3 and 5 year survival, early deaths and mid-term type I endoleaks. CONCLUSIONS: An interactive model is available, which can assist vascular surgeons to evaluate the expected outcomes for a particular patient undergoing EVAR. The validated model is useful for counselling and pre-operative decision making.
Authors: Chieh Yang Koo; Joseph A Hyder; Jonathan P Wanderer; Matthias Eikermann; Satya Krishna Ramachandran Journal: World J Surg Date: 2015-01 Impact factor: 3.352
Authors: Alan Karthikesalingam; Sandeep S Bahia; Shaneel R Patel; Bilal Azhar; Dan Jackson; Lynne Cresswell; Robert J Hinchliffe; Peter J E Holt; Matt M Thompson Journal: Kidney Int Date: 2014-08-20 Impact factor: 10.612
Authors: George A Antoniou; Marc L Schermerhorn; Thomas L Forbes; Vincent Cheng; Stavros A Antoniou; Jonathan Golledge; Hence J M Verhagen; Francesco Torella Journal: BMJ Open Date: 2022-04-25 Impact factor: 2.692