OBJECTIVES: Aortic valve replacement (AVR) is accepted as the standard treatment for severe symptomatic aortic valve stenosis and regurgitation. As novel treatments are introduced for patients at high risk for conventional surgery, it is important to have models that accurately predict procedural risk. The aim of this study was to develop and validate a risk-stratification model to predict in-hospital risk of death for patients undergoing AVR and to compare the model with existing algorithms. METHODS: We reviewed data from the Central Cardiac Adult Database, which holds prospectively collected clinical information on all adult patients undergoing cardiac surgery in National Health Service (NHS) hospitals and some private providers in the UK and Ireland. We included all the patients undergoing AVR with or without coronary artery bypass grafting. The study population consists of 55 157 patients undergoing surgery between 1 April 2001 and 31 March 2009. The model was built using data from April 2001 to March 2008 and validated using data from patients undergoing surgery from April 2008 to March 2009. The model was compared against the additive and logistic EuroSCORE models and a valve-specific risk-prediction model. RESULTS: The final multivariable model includes items describing cardiovascular risk status and procedural factors. Applying the model to the independent validation dataset provided a c-statistic (index of rank correlation) of 0.791, which was substantially better than that achieved by previously developed risk models in Europe, and significantly improved risk prediction in higher-risk patients. CONCLUSIONS: We have produced an accurate risk model to predict outcome following AVR surgery. It will be of use for patient selection and informed consent, and of particular interest in defining those patients at high risk who may benefit from novel approaches to AVR.
OBJECTIVES: Aortic valve replacement (AVR) is accepted as the standard treatment for severe symptomatic aortic valve stenosis and regurgitation. As novel treatments are introduced for patients at high risk for conventional surgery, it is important to have models that accurately predict procedural risk. The aim of this study was to develop and validate a risk-stratification model to predict in-hospital risk of death for patients undergoing AVR and to compare the model with existing algorithms. METHODS: We reviewed data from the Central Cardiac Adult Database, which holds prospectively collected clinical information on all adult patients undergoing cardiac surgery in National Health Service (NHS) hospitals and some private providers in the UK and Ireland. We included all the patients undergoing AVR with or without coronary artery bypass grafting. The study population consists of 55 157 patients undergoing surgery between 1 April 2001 and 31 March 2009. The model was built using data from April 2001 to March 2008 and validated using data from patients undergoing surgery from April 2008 to March 2009. The model was compared against the additive and logistic EuroSCORE models and a valve-specific risk-prediction model. RESULTS: The final multivariable model includes items describing cardiovascular risk status and procedural factors. Applying the model to the independent validation dataset provided a c-statistic (index of rank correlation) of 0.791, which was substantially better than that achieved by previously developed risk models in Europe, and significantly improved risk prediction in higher-risk patients. CONCLUSIONS: We have produced an accurate risk model to predict outcome following AVR surgery. It will be of use for patient selection and informed consent, and of particular interest in defining those patients at high risk who may benefit from novel approaches to AVR.
Authors: Daniel Paul Fudulu; Dan Mihai Dorobantu; Mansour Taghavi Azar Sharabiani; Gianni Davide Angelini; Massimo Caputo; Andrew John Parry; Serban Constantin Stoica Journal: Open Heart Date: 2015-09-04