Multivariable Prediction Of In-hospital Mortality Associated With Aortic Valve Replacement In The Adult

【Objective】Over the 20-year study period, we analyzed a consecutive series of aortic valve replacement in adults (1) to determine implications for patients referred for AVR (2) to analyze the predictive value of the European system for cardiac operative risk evaluation score (EuroSCORE) and the Society of Thoracic Surgeons predicted risk of mortality (STS-PROM) in adults undergoing aortic valve replacement (AVR) (3) and, to determine predictors of in-hospital mortality after aortic valve replacement.【Methods】1,Aortic valve replacement in adults: a retrospective study of 927 cases.(1) The study population consisted of patients 18 years old or older who underwent primary aortic valve procedure at Changhai hospital between 1999 and 2008. Patients were excluded if they underwent combined valve replacement or aortic valve replacement for aortic dissection. Patients with an additional surgical procedure, including concomitant coronary artery bypass grafting, correction of congenital heart disease, thoracic aortic surgery and valvuloplasty for mitral or tricuspid regurgitation caused by aortic valve disease, were also included.(2) The clinical outcome considered was in-hospital mortality, defined as the patient's status at discharge after the operation.2,Validation of the EuroSCORE and the STS-PROM in adult patients undergoing aortic valve replacement.(1) We carried out a retrospective statistical analysis on adult patients undergoing AVR between 1999 and 2008 according to the definitions of the EuroSCORE or STS-PROM. Patients with concomitant coronary artery bypass grafting were also included. Excluded from this study were patients having surgery for congenital heart defects, aneurysm of thoracic aorta and atrial fibrillation. Operative mortality was defined as death before discharge from the hospital (in-hospital mortality).(2) The mortality risk calculation of EuroSCORE and STS-PROM for aortic valve procedures was performed by the online available EuroSCORE or STS score calculator. Based on the additive EuroSCORE risk calculation, patients were divided into low-risk, medium-risk and high-risk groups.(3) The valuation of three different algorithms depended on the assessment of two features: calibration and discrimination. A comparison of observed and predicted mortality rates was also performed.3,Multivariable prediction of in-hospital mortality associated with aortic valve replacement in the adult.(1) Data was collected on adult patients undergoing aortic valve replacement or aortic valve replacement plus coronary artery bypass surgery between 1995 and 2009 from registry of our institute. Excluded from this study were patients having combined valve replacement, patients with mitral or tricuspid valvoplasty, patients with surgical correction of congenital heart disease, and patients undergoing surgery of the thoracic aorta. The end point wasin-hospital mortality. (2) The univariate association between risk factors and in-hospital mortality was assessed with a univariate analysis.(3) All variables significant in the univariate analysis were subsequently entered into a multivariable logistic regression analysis.【Results】1,Aortic valve replacement in adults: a retrospective study of 927 cases.(1) During the 20-year study period, a total of 927 patients were identified as having aortic valve replacement, 92% of which resenting with a variety of different symptoms. Both the number of AVRs and the age of patients increased,while course of the disease became shorter. The major cause of the aortic valve disease included congenital valve, degenerative and rheumatic change. There was a shift in the causes of aortic valve disease. Patients with aortic regurgitation (AR) were younger compare to other groups (P<0.001). There was no age difference between AS and AS+AR group.(2) The proportion of AVR combined with other procedure increased between 2000 and 2009. For isolated aortic valve surgery, the CPB time and aortic clamping time were significantly shorter than those of before (P <0.05). Aortic valve replacement using mechanical valve accounted for 91.3% of aortic valve surgical procedure. For the aortic position, the most commonly used valve were Carbomedics series (54%). Among the various sizes of aortic valve prostheses, the most commonly used valves were 23mm and 25mm size for males and 21mm, 23mm and 25mm size for females. (3) During the Last Decade, duration of postoperative ventilation (P <0.05), time in the intensive care unit (ICU) (P <0.001) and length of postoperative hospital stay (P <0.001) reduced significantly compare with those of the first decade. The overall in-hospital mortality was 4.7% (44 of 927 patients). No gender difference was found in operative mortality. The mortality have greatly decreased in the past 15 years,and in-hospital mortality of the last 5 years (2005-2009) was lower than that of the first 5 years (1990-1994)(P<0.05). The mortality in patients over 60 years was higher than that in younger (≤60 years) patients (8.8% vs. 3.8%, P <0.01).2,Validation of the EuroSCORE and the STS-PROM in adult patients undergoing aortic valve replacement:(1) A total of 521 patients were identified as having undergone aortic valve replacement. In-hospital mortality was 4% overall. The expected mortality for the additive, logistic EuroSCORE and the STS-PROM was 3.36%, 2.82% and 1.25%, respectively. The observed to expected ratio was 1.2 for additive EuroSCORE,1.43 for logistic EuroSCORE and 3.23 for STS-PROM. The STS-PROM underpredicted observed mortality significantly (P<0.0001). The logistic EuroSCORE showed a tendency to underpredict observed mortality. However, additive EuroSCORE was close to actual mortality.(2) There were 193, 275 and 53 patients classified as low-, medium- and high-risk group according to the additive EuroSCORE score, and the observed mortality was 0.5%, 4.7% and 13.2%, respectively. The STS-PROM underpredicted observed mortality in medium- and high-risk group, significantly (P<0.01). The logistic EuroSCORE underpredicted observed mortality in the medium-risk subgroup, significantly (P<0.05). EuroSCORE show a tendency to underpredict observed mortality in the high-risk subgroup with the observed to expected radio of 1.84 for additive EuroSCORE and 1.46 for logistic EuroSCORE.(3) The discriminative power of models for the entire cohort with the area under the ROC curve of 0.727 for additive EuroSCORE,0.753 for logistic EuroSCORE and 0.753 for STS-PROM. The EuroSCORE in three subgroups showed poor discrimination in predicting mortality as well as the STS-PROM did in the medium- and high-risk subgroups.3,Multivariable prediction of in-hospital mortality associated with aortic valve replacement in the adult:(1) A total of 693 patients were identified as having undergone aortic valve replacement. In-hospital mortality was 4% overall. The leading causes of death were low cardiac output syndrome and severe arrhythmia, which accounted for 53.3% of deaths. No gender difference was found in operative mortality.(2) Candidate predictors significant in the univariate analysis were age, time-course of disease, diabetes, chronic obstructive pulmonary disease, creatinine,atrial fibrillation, left ventricular ejection fraction, shortening fraction, interventricular septal thickness, left ventricular posterior wall thickness, relative wall thickness, left atrial volume, right atrial volume, aortic stenosis, aortic insufficiency, abnormal coronary angiography, NYHA functional class, preoperative myocardial infarction, concomitant CABG surgery, cardiopulmonary bypass (CPB) time, cross-clamp time and assisted CPB time.(3) Atrial fibrillation, left ventricular posterior wall thickness and concomitant CABG surgery were confirmed to be independent predictors of in-hospital mortality.【Conclusions】1,During the 20-year study period, with improved surgical techniques , better intensive care treatment and different protocols of myocardial protection, more and more patients referred for aortic valve replacement. The in-hospital mortality decreased to a level lower than that of before.2,Both the EuroSCORE and the STS-PROM give an imprecise prediction for individual operative risk in patients undergoing aortic valve replacement in our study. These algorithms seem unsuitable to identify a high-risk patient population undergoing isolated AVR. It is necessary to construct a risk stratification model for valve surgery according to the profiles of Chinese patients.3,By multivariable logistic regression analysis, some associated Risk Factors for AVR were confirmed to be independent predictors of in-hospital mortality, which could be used to identify patients at high risk of postoperative morbidity and mortality.