| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2004;77:1966-1977
© 2004 The Society of Thoracic Surgeons
a Department of Surgery, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA
b Department of Surgery, Catholic Medical Center, Manchester, New Hampshire, USA
c Department of Surgery, Concord Hospital, Concord, New Hampshire, USA
d Department of Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire, USA
e Eastern Maine Medical Center, Bangor, Maine, USA
f Department of Surgery, Fletcher Allen Health Care, Burlington, Vermont, USA
g Department of Surgery, Maine Medical Center, Portland, Maine, USA
h Department of Surgery, Portsmouth Regional Hospital, Portsmouth, New Hampshire, USA
i Dartmouth Medical School, Hanover, New Hampshire, USA
Accepted for publication December 2, 2003.
* Address reprint requests to Dr O'Connor, Dartmouth Medical School, One Medical Center Dr, Lebanon, NH 03756, USA
e-mail: gerald.t.oconnor{at}dartmouth.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
METHODS: Data from eight Northern New England Medical Centers in the period January 1991 through December 2001 were analyzed on 8,943 heart valve surgery patients aged 30 years and older. There were 5,793 cases of aortic valve replacement and 3,150 cases of mitral valve surgery (repair or replacement). Logistic regression was used to examine the relationship between risk factors and in-hospital mortality.
RESULTS: In the multivariable analysis, 11 variables in the aortic model (older age, lower body surface area, prior cardiac operation, elevated creatinine, prior stroke, New York Heart Association [NYHA] class IV, congestive heart failure [CHF], atrial fibrillation, acuity, year of surgery, and concomitant coronary artery bypass grafting) and 10 variables in the mitral model (female sex, older age, diabetes, coronary artery disease, prior cerebrovascular accident, elevated creatinine, NYHA class IV, CHF, acuity, and valve replacement) remained independent predictors of the outcome. The mathematical models were highly significant predictors of the outcome, in-hospital mortality, and the results are in general agreement with those of others. The area under the receiver operating characteristic curve for the aortic model was 0.75 (95% confidence interval [CI], 0.72 to 0.77), and for the mitral model, 0.79 (95% CI, 0.76 to 0.81). The goodness-of-fit statistic for the aortic model was 
2 [8 df] = 11.88, p = 0.157, and for the mitral model it was 2 [8 df] = 5.45, p = 0.708.
CONCLUSIONS: We present results and methods for use in day-to-day practice to calculate patient-specific in-hospital mortality after aortic and mitral valve surgery, by the logistic equation for each model or a simple scoring system with a look-up table for mortality rate.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
Great progress has been made in identifying and defining risk factors and reaching consensus with regard to their use in risk adjusting rates of adverse outcomes with coronary artery bypass surgery (CABG) [1–4]. While much has been written about independent risk factors for mortality after heart valve surgery, the studies frequently involve relatively limited numbers of cases with analysis of both short-term and long-term outcomes [5–10]. For a number of reasons, research regarding risk prediction for valve surgery has lagged behind that for CABG. There are fewer valve surgery cases than coronary bypass cases, requiring more time to amass adequate numbers of patients to evaluate outcomes. This problem is compounded by changes in patient characteristics, surgical techniques, and outcomes over time [11]. Mortality rates and risk factors also vary according to valve location and procedure.
Recently, however, studies based on large data sets from the New York State, Society for Thoracic Surgery, and EuroSCORE cardiac surgery registries have identified important risk factors and presented methods for risk modeling with predictors of short-term mortality in patients undergoing heart valve surgery as the primary procedure [12–14]. We report the development of multivariable prediction rules for in-hospital mortality in patients undergoing aortic and mitral valve surgery and demonstrate a method for their practical use in clinical practice.
|
Material and methods |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
Data in the registry have been previously described [11]. Priority at operation was defined as follows: (1) emergency, the patient's clinical condition required operation within hours to prevent morbidity or death; (2) urgent, the clinical condition required operation within days before discharge from the hospital; (3) elective, the clinical condition required operation, but operation could be done on another admission. Preoperative and operative variables were selected for analysis based upon review of the literature, experience of the clinicians in the research group, and availability and completeness of data in the registry (Table 1).
|
Two statistical models were created, one for replacement of the aortic valve with or without CABG and one for mitral valve procedures with either repair or replacement. Risk factors with a p value less than 0.20 in the univariate analysis were included in the initial multivariable model. Bootstrap bagging with automated forward stepwise selection in 200 samples of 75% from the original data set was used to confirm our assumption about the importance of the risk factors in the study. The frequency of occurrence of variables was determined and used as an indicator of reliability [17, 18]. The likelihood ratio 2 test (2 LR) was used to compare nested logistic regression models. Risk factors were retained in the final models if the p value was less than 0.05.
The area under the receiver operating characteristic curve (ROC) was used as a measure of model discrimination [19]. The statistical models were internally validated using the technique of bootstrap resampling [20]. Each entire set of aortic and mitral cases was used in turn to develop a model specific to each valve site (an aortic and a mitral model). The bootstrap sampling technique was run for each model (aortic or mitral) by drawing at random with replacement 100 samples of 100% of the cases to calculate an individual sample ROC value and then the mean and standard error of the mean with 95% confidence intervals (95% CI) for all 100 ROC values. Observed and predicted numbers of deaths by decile of predicted risk were compared to assess calibration of the prediction equation. Hosmer-Lemeshow goodness-of-fit statistics were calculated for both models [21].
The relative contribution of groups of variables to the prediction of in-hospital mortality was calculated. Groups of variables were systematically dropped from each model in turn. This difference between the total LR 2 of the full model and the 2 of the smaller model was recorded for each group and considered the dropped group's contribution to the prediction of mortality in the model [1]. Each individual group contribution was then compared to the sum of all group contributions to identify the relative contribution of each group to the prediction of in-hospital mortality.
A clinical risk assessment tool was developed from the final regression models to predict the probability of in-hospital mortality after either aortic or mitral valve surgery by entering the logistic equation into an Excel (Microsoft Corp, Redmond,WA) spread sheet. This is our preferred method of estimating the risk for mortality. An alternative method is a scorecard. We arranged our risk factor indicator variables in the logistic model to have positive coefficients with resultant odds ratios of one or greater. Each variable in the model was assigned risk-of-death points by rounding off its OR to the nearest 0.5. Totaling these points (weights) across those risk factors applicable to each patient resulted in a summary score reflective of the patient's unique risk of in-hospital mortality. From the logistic equation, a predicted probability of death was also calculated for each patient. The possible summary scores for all patients were collapsed into 20 groups. Then the mean predicted probability of death was calculated for each summary score group from the individual patient estimates (logistic equation). For both the aortic and mitral models, the mortality score was adjusted to the mortality rates for the period, 1999 to 2001, to reflect current case mix and standards of care.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
|
|
2 LR [19 df] = 269.91, p < 0.001). Independent predictors of in-hospital mortality included older age, smaller BSA, prior cardiac operation, increased creatinine level, NYHA class IV, CHF and atrial fibrillation, urgent or emergent priority, concomitant CABG, and earlier year of operation. All independent predictor variables met a bootstrap reliability of 50% or greater. The multivariable model was then tested against the full model including all significant univariate predictors that had been previously excluded. The reintroduction of these dropped variables did not significantly improve the model (2 LR [6 df] = 4.07, p = 0.667). The area under the model ROC curve was 0.75, and the calculated average ROC from the bootstrap resampling was 0.75 (95% CL = 0.72 to 0.77). The aortic model thus showed good ability to discriminate between those dying and surviving in-hospital after surgery (Fig 1).
To test model calibration we compared observed and predicted death rates within increasing deciles of risk. The correlation between observed and expected deaths was 0.98. The Hosmer-Lemeshow goodness-of-fit statistic was 2 [8 df] = 11.88, p = 0.157 (Fig 2).
All clinically important first-order interactions were tested. None was found to be statistically significant. Table 4
summarizes the variables used in the aortic model, their regression coefficients, and associated p values.
|
|
|
2 LR [16 df] = 300.10, p < 0.001). Independent predictors of in-hospital mortality in the mitral model included female gender, older age, diabetes, CAD, prior stroke, increased creatinine level, NYHA class IV, CHF, urgent or emergent priority, and replacement rather than repair of the mitral valve. These variables also met the bootstrap reliability criterion of 50% or greater. Reintroducing all significant univariate predictors previously dropped contributed little to the mitral model (chi2 LR [9 df] = 6.59, p = 0.680). In Figure 1, the area under the ROC curve for the mitral model was 0.79 with a bootstrap calculated average ROC of 0.79 (95% CL = 0.76 to 0.81). The correlation between observed and expected deaths was 0.99, and the Hosmer-Lemeshow goodness-of-fit statistic for the mitral model was 2 [8 df] = 5.45, p = 0.704, indicating little departure from perfect fit (Fig 2). There were no statistically significant first-order interactions. Table 5
summarizes the variables used in the mitral prediction model, their regression coefficients and associated p values. While not a statistically significant risk factor, year of operation is included for adjustment purposes.
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
An important aspect of this study is that both models developed included only patient characteristics and history, comorbid conditions, and patient clinical status variables immediately available to clinicians before surgery. Knowledge of a patient's predicted risk of mortality can serve to augment clinical judgment at the time of patient counseling. In addition to presenting mortality rates and relative risk (OR), we show the relative contribution of these factors to in-hospital mortality in Figure 3 [1]. We believe the effect of these logical clinical groupings is to stimulate a greater awareness among clinicians of the significance of specific risk factors in individual patients who face heart valve surgery.
Both of our models lack a certain degree of precision, but compare favorably to the models of others (see below). There remains approximately 0.25 (aortic) and 0.21 (mitral) of the area over the ROC curves that is unexplained. First, we have limited our predicting variables to those known before the operation occurred, so that these variables alone could hardly provide a perfect prediction for in-hospital mortality. As an example, echocardiographic and catheterization data related to the severity of valve pathology and ascending aorta atherosclerotic disease are not captured in our registry. These data might have been surrogates for a more technically challenging procedure with a negative effect on the outcome. Second, the models did not include variables reflecting variations in surgical technique, intraoperative or postoperative care by center, or individual surgeon (eg, myocardial protection techniques). As a practical matter, it would be impossible to include these in a model used for preoperative prediction of outcome, though their inclusion might improve prediction. Third, we did not have reliable data for the clinical diagnosis of the valve pathology. In the mitral model, knowing whether the pathologic cause of mitral regurgitation was degenerative or ischemic-related may have improved discrimination.
Finally, our valve registry may not include some variables that would help to predict in-hospital mortality. For example, ejection fraction alone may not be a sophisticated enough measure for the degree of myocardial pathophysiologic and dysfunctional change in valvular heart disease. Measures of increased left ventricular mass or severity of left ventricular hypertrophy may be associated with difficulty in myocardial protection and resultant increased morbidity and mortality after operation, particularly in aortic cases. Other measures of left ventricular volume such as end systolic and end diastolic dimensions may have more effect on mortality in the longer term. Except for left ventricular ejection fraction, we were unable to measure these other factors reliably during the time frame of this study.
Prior reports of clinical prediction rules for heart valve surgery in studies with a large number of cases have appeared in the literature. Using logistic regression techniques applied to data from the Society for Thoracic Surgeons National Database, Jamieson and colleagues [13] reported the development of risk strata for postoperative mortality after cardiac valve replacement. The analysis of the 1986 to 1995 series of 86,580 cases from participating institutions included six models for different combinations of procedures performed. The study examined 51 preoperative risk factors, of which 30 were significant independent predictors in the various logistic models. In general, age, emergency status, renal failure, and NYHA class were the most important predictors for all models. The addition of CABG increased the risk of in-hospital death for all subset models. The study did not include valve repair procedures and therefore could not provide information about repair versus replacement as a risk factor, particularly in mitral valve surgery. The contributions of individual centers could be used for risk-adjusted comparisons to a national standard, but no coefficients were reported for use in predicting the outcome for individual patients, nor were the areas under the ROC curves reported.
Hannan and colleagues [12] published a report in 2000 on predictors of mortality for patients undergoing cardiac valve replacements in New York State. The intent was to contrast the mortality rates and significant risk factors among fairly homogeneous groups of valve patients. The series of 14,190 cases between 1995 and 1997 was therefore stratified into six groups according to whether aortic, mitral, or multiple valve procedures were performed, with and without CABG. Valve repair operations were included in the same category as replacements. Shock carried the highest risk (nearly nine times) of mortality in both aortic and mitral operations with and without CABG, but occurred in only 0.50% to 0.60% of the aortic cases and 1.51% to 4.14% of mitral cases. Salvage status, emergency status, and NYHA class were not available in the study. Congestive heart failure was defined as NYHA class III or IV. Among 18 preoperative independent risk factors identified, age, hemodynamic instability, shock, congestive heart failure in the same admission, diabetes, renal dialysis, EF less than 30%, extensively calcified aorta, and female gender were the most important, but varied from model to model, particularly if CABG was also performed. Neither severity of stenosis nor of regurgitation was an independent predictor of death. The C statistic (area under the ROC curve) varied from 0.718 to 0.823 for the six different models, but no coefficients for the logistic equation were reported to permit prediction in individual patients.
Roques and colleagues [14], using September to December 1995 data on 5,672 patients having heart valve surgery in 132 European centers, published information regarding a method, the EuroSCORE, for predicting early mortality. The purpose of the study was to provide a tool for epidemiologic assessment of individual surgeon, institution, or country results. Age, creatinine, previous heart surgery, left ventricular function, CHF, pulmonary hypertension, endocarditis, emergency procedure, multiple valve or tricuspid procedures, and combined CABG or thoracic surgery were predictive factors. The discriminatory ability of the system was considered good with an area under the ROC curve as high as 0.75, and the calibration over the spectrum of risk was satisfactory as measured by the Hosmer-Lemeshow goodness-of-fit test.
Compared to these other studies, we found many of the same independent predictors. Older age, smaller size, and urgent or emergent priority at operation were the greatest contributors to mortality in the aortic model, and older age, sex, and urgent and emergent presentation the most important in the mitral model. A history of past or current CHF and NYHA class IV remained in both the aortic and mitral models in our study. This may be due to a difference in definition between the two variables. The CHF definition may include more the consideration of treatment and NHYA class more the activity level of the patient. In our original data, NYHA class had a fifth category defined as the patient being in shock at the time of operation. These patients are likely comparable to those with shock in the Hannan study. Because of the low numbers of these patients, they were included in the traditional class IV category. As a consequence, some of the significance of NYHA class IV reflects the influence of a small number of highly unstable patients on the risk of dying. As in the Hannan study, severity of stenosis or regurgitation of the valve was not statistically significant. Our aortic model used concomitant CABG as a predictor, and the mitral model did not since, in the face of other variables in the model, concomitant CABG was not an independent predictor. In contrast to the other studies, we used mitral procedure (repair vs replacement) as a predictor variable in the mitral model. Since we had excluded multiple valve procedures as well as tricuspid and pulmonic valve operations because of small numbers, we decreased the complexity of the prediction rule to two models. These cases of aortic and mitral valve surgery constitute large numbers of heart valve procedures performed in our region (78%) and the procedures that physicians and patients together must consider in the majority of clinical situations.
Future work should evaluate intermediate and long-term mortality with echocardiographic measures for altered left ventricular geometry, volume and severity of hypertrophy, cited by others as risk factors not only for early mortality after heart valve surgery, but also for long-term survival [23–25]. Collection of data for the analysis of the pathologic cause of mitral regurgitation is currently under consideration, together with adverse in-hospital outcomes other than in-hospital mortality, such as postoperative neurologic dysfunction, mediastinal infection, and bleeding requiring reoperation.
The goals of this study were several: (1) to foster greater awareness among a diverse group of clinicians in our region of the risk factors for an early fatal outcome after valve surgery; (2) to inform patient decision making; and (3) to change processes of care to neutralize these risk factors. Opinions about whether a surgical procedure is low, medium, or high risk may vary from clinician to clinician. We believe that the logistic equation to estimate the risk of in-hospital mortality should be used whenever possible to eliminate some of this subjectivity. The scorecard approach is a less accurate, but nevertheless useful and convenient alternative. Applicability of these prediction models to future cases is the true test of their validity. Over time these rules can be validated in practice and revised appropriately.
|
Conclusion |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterial and methodsResultsCommentConclusionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. C. Herrmann Transcatheter Aortic Valve Implantation: Past, Present, and Future Circ Cardiovasc Intervent, December 1, 2008; 1(3): 159 - 160. [Full Text] [PDF]
|
|
|
|
 |
|
 2006 WRITING COMMITTEE MEMBERS, R. O. Bonow, B. A. Carabello, K. Chatterjee, A. C. de Leon Jr, D. P. Faxon, M. D. Freed, W. H. Gaasch, B. W. Lytle, R. A. Nishimura, et al. 2008 Focused Update Incorporated Into the ACC/AHA 2006 Guidelines for the Management of Patients With Valvular Heart Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1998 Guidelines for the Management of Patients With Valvular Heart Disease): Endorsed by the Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons Circulation, October 7, 2008; 118(15): e523 - e661. [Full Text] [PDF]
|
|
|
|
 |
|
 R. O. Bonow, B. A. Carabello, K. Chatterjee, A. C. de Leon Jr, D. P. Faxon, M. D. Freed, W. H. Gaasch, B. W. Lytle, R. A. Nishimura, P. T. O'Gara, et al. 2008 Focused Update Incorporated Into the ACC/AHA 2006 Guidelines for the Management of Patients With Valvular Heart Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1998 Guidelines for the Management of Patients With Valvular Heart Disease) Endorsed by the Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, and Society of Thoracic Surgeons J. Am. Coll. Cardiol., September 23, 2008; 52(13): e1 - e142. [Full Text] [PDF]
|
|
|
|
 |
|
 J. Granton and D. Cheng Risk Stratification Models for Cardiac Surgery Seminars in Cardiothoracic and Vascular Anesthesia, September 1, 2008; 12(3): 167 - 174. [Abstract] [PDF]
|
|
|
|
 |
|
 M. van Gameren, A. P. Kappetein, E. W. Steyerberg, A. C. Venema, E. A.J. Berenschot, E. L. Hannan, A. J.J.C. Bogers, and J. J.M. Takkenberg Do We Need Separate Risk Stratification Models for Hospital Mortality After Heart Valve Surgery? Ann. Thorac. Surg., March 1, 2008; 85(3): 921 - 930. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. A. Ferraris, F. H. Edwards, D. M. Shahian, and S. P. Ferraris Risk Stratification and Comorbidity Card. Surg. Adult, January 1, 2008; 3(2008): 199 - 246. [Full Text]
|
|
|
|
|
|
T. Gudbjartsson, T. Absi, and S. Aranki Mitral Valve Replacement Card. Surg. Adult, January 1, 2008; 3(2008): 1031 - 1068. [Full Text]
|
|
|
|
 |
|
 Y. S. Tjang, Y. van Hees, R. Korfer, D. E. Grobbee, and G. J.M.G. van der Heijden Predictors of mortality after aortic valve replacement Eur. J. Cardiothorac. Surg., September 1, 2007; 32(3): 469 - 474. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.-A. Rey Meyer, L. K. von Segesser, M. Hurni, F. Stumpe, K. Eisa, and P. Ruchat Long-term outcome after mitral valve repair: a risk factor analysis Eur. J. Cardiothorac. Surg., August 1, 2007; 32(2): 301 - 307. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. M. Shahian, F. H. Edwards, V. A. Ferraris, C. K. Haan, J. B. Rich, S.-L. T. Normand, E. R. DeLong, S. M. O'Brien, C. M. Shewan, R. S. Dokholyan, et al. Quality Measurement in Adult Cardiac Surgery: Part 1--Conceptual Framework and Measure Selection Ann. Thorac. Surg., April 1, 2007; 83(4_Supplement): S3 - S12. [Full Text] [PDF]
|
|
|
|
|
|
E. L. Hannan, C. Wu, E. V. Bennett, R. E. Carlson, A. T. Culliford, J. P. Gold, R. S.D. Higgins, C. R. Smith, and R. H. Jones Risk Index for Predicting In-Hospital Mortality for Cardiac Valve Surgery Ann. Thorac. Surg., March 1, 2007; 83(3): 921 - 929. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Florath, A. Albert, W. Hassanein, B. Arnrich, U. Rosendahl, I. C. Ennker, and J. Ennker Current determinants of 30-day and 3-month mortality in over 2000 aortic valve replacements: impact of routine laboratory parameters Eur. J. Cardiothorac. Surg., November 1, 2006; 30(5): 716 - 721. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. O. Bonow, B. A. Carabello, K. Chatterjee, A. C. de Leon Jr, D. P. Faxon, M. D. Freed, W. H. Gaasch, B. W. Lytle, R. A. Nishimura, P. T. O'Gara, et al. ACC/AHA 2006 Guidelines for the Management of Patients With Valvular Heart Disease: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1998 Guidelines for the Management of Patients With Valvular Heart Disease) Developed in Collaboration With the Society of Cardiovascular Anesthesiologists Endorsed by the Society for Cardiovascular Angiography and Interventions and the Society of Thoracic Surgeons J. Am. Coll. Cardiol., August 1, 2006; 48(3): e1 - e148. [Full Text] [PDF]
|
|
|
|
|
|
R. O. Bonow, B. A. Carabello, K. Chatterjee, A. C. de Leon Jr, D. P. Faxon, M. D. Freed, W. H. Gaasch, B. W. Lytle, R. A. Nishimura, P. T. O'Gara, et al. ACC/AHA 2006 Practice Guidelines for the Management of Patients With Valvular Heart Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 1998 Guidelines for the Management of Patients With Valvular Heart Disease) Developed in Collaboration With the Society of Cardiovascular Anesthesiologists Endorsed by the Society for Cardiovascular Angiography and Interventions and the Society of Thoracic Surgeons J. Am. Coll. Cardiol., August 1, 2006; 48(3): 598 - 675. [Full Text] [PDF]
|
|
|
|
 |
|
 J. Nilsson, M. Ohlsson, L. Thulin, P. Hoglund, S. A.M. Nashef, and J. Brandt Risk factor identification and mortality prediction in cardiac surgery using artificial neural networks J. Thorac. Cardiovasc. Surg., July 1, 2006; 132(1): 12 - 19. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. B. Vanky, E. Hakanson, E. Tamas, and R. Svedjeholm Risk Factors for Postoperative Heart Failure in Patients Operated on for Aortic Stenosis Ann. Thorac. Surg., April 1, 2006; 81(4): 1297 - 1304. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. L. Grunkemeier and A. P. Furnary Mandatory Database Participation: Risky Business? Ann. Thorac. Surg., September 1, 2005; 80(3): 799 - 801. [Full Text] [PDF]
|
|
|
|
|
|
E. R. Nowicki What is the Future of Mortality Prediction Models in Heart Valve Surgery? Ann. Thorac. Surg., August 1, 2005; 80(2): 396 - 398. [Full Text] [PDF]
|
|
|
|
|
|
R. Jin, G. L. Grunkemeier, A. Starr, and Providence Health System Cardiovascular Study Grou Validation and Refinement of Mortality Risk Models for Heart Valve Surgery Ann. Thorac. Surg., August 1, 2005; 80(2): 471 - 479. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Jin, G. L. Grunkemeier, and For the Providence Health System Cardiovascular St Additive vs. logistic risk models for cardiac surgery mortality Eur. J. Cardiothorac. Surg., August 1, 2005; 28(2): 240 - 243. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Ambler, R. Z. Omar, P. Royston, R. Kinsman, B. E. Keogh, and K. M. Taylor Generic, Simple Risk Stratification Model for Heart Valve Surgery Circulation, July 12, 2005; 112(2): 224 - 231. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Nagendran, C. Norris, A. Maitland, A. Koshal, and D. B. Ross Is mitral valve surgery safe in octogenarians? Eur. J. Cardiothorac. Surg., July 1, 2005; 28(1): 83 - 87. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. K. Toumpoulis, C. E. Anagnostopoulos, S. K. Toumpoulis, J. J. DeRose Jr, and D. G. Swistel EuroSCORE Predicts Long-Term Mortality After Heart Valve Surgery Ann. Thorac. Surg., June 1, 2005; 79(6): 1902 - 1908. [Abstract] [Full Text] [PDF]
|
|
| |||||||
