EJCTS Click here to go to Siemens website
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to Personal Folders
Right arrow Download to citation manager
Right arrow Author home page(s):
Luc Noyez
Rene M.H.J. Brouwer
Right arrow Permission Requests
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by van Eck, F. M.
Right arrow Articles by Brouwer, R. M.H.J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by van Eck, F. M.
Right arrow Articles by Brouwer, R. M.H.J.
Related Collections
Right arrow Coronary disease

Eur J Cardiothorac Surg 2002;21:1031-1036
© 2002 Elsevier Science NL

Preoperative prediction of early mortality in redocoronary artery surgery

Frans M. van Eck, Luc Noyez*, Freek W.A. Verheugt, Rene M.H.J. Brouwer

Department of Thoracic and Cardiac Surgery -414-, Heartcenter, University Nijmegen Medical Center, St. Radboud, 6500 HB Nijmegen, The Netherlands

Received 17 December 2001; received in revised form 18 February 2002; accepted 15 March 2002.

* Corresponding author. Tel.: +31-24-3614744; fax: +31-24-3540129
e-mail: l.noyez{at}thorax.umcn.nl


   Abstract
Top
 Abstract
1. Introduction
 2. Materials and methods
 3. Results
 4. Discussion
 Appendix A. Example of...
 References
 
Objectives: Construct a predictive model for early mortality in coronary reoperations (RECABG). Methods: Five hundred and forty one RECABG (1987–1998) were studied by univariate and multivariate analysis. Stepwise selective procedure (p<0.05) was used to identify a subset of variables with prognostic value for early mortality. This subset was used to calculate a prognostic score ‘S’ and a predicted probability ‘P’ for early mortality, P=1/1+e-S. Sensitivity analysis was used for evaluation. Results: The best predictive variables for early mortality were diabetes, vascular-, lung-disease, a myocardial infarction between the primary and the RECABG, acute- and emergency operation and the operative period. The prognostic accuracy (receiver operating characteristics curve (ROC) area) was 80%. Observed probabilities compare well with the predicted probabilities, and patients were classified in low risk (5%), intermediate risk (15%), high risk (30%) and very high risk (40%). A predicted probability of >=0.40 was used as cut-off point for the prognostic test. The specificity of this test was 97%, sensitivity 33%, predictive value of a positive test 63% and 90% for a negative test. Conclusions: The results show that individual patients presented for RECABG, can be stratified according to their early mortality risk. This information can be used to inform the patient, and also to discus the opportunity of the RECABG.

Key Words: Reoperation • Early mortality • Prediction • Myocardial revascularization


   1. Introduction
Top
 Abstract
 1. Introduction
2. Materials and methods
 3. Results
 4. Discussion
 Appendix A. Example of...
 References
 
Mortality of coronary reoperations (RECABG) is still higher than in primary myocardial revascularization [1–6]. As in all surgical procedures, to improve the results, progress has to be made in the preoperative, perioperative and postoperative phase. A better, evidence-based, patient selection can be a first step in the improvement of the results of RECABG. The intention of this study is to construct a predictive model of early mortality of RECABG, using preoperative variables. In the study, early mortality is the endpoint. Early mortality, 6 months postoperative, is related to patient variables and the surgical procedure [7], and is therefore a more honourable endpoint than surgery, hospital stay or 30-day mortality, which are used as endpoints in most other reports.


   2. Materials and methods
Top
 Abstract
 1. Introduction
 2. Materials and methods
3. Results
 4. Discussion
 Appendix A. Example of...
 References
 
2.1. Patients
With the aid of our database, Coronary Surgery Database Radboud Hospital (CORRAD), a registry that stores pre-, peri-, postoperative and follow-up data of all patients undergoing isolated coronary bypass grafting, we identified a series of 541 patients undergoing a first coronary reoperation (RECABG) from January 1987 to December 1998 at the UMC St. Radboud Nijmegen. Only patients with an isolated myocardial revascularization at the primary operation were included in this study.

The studied variables and their definition are presented in Table 1.


View this table:
[in this window]
[in a new window]
  		Table 1. Unifactor risk analysis for early mortality

 
2.2. Surgical technique
Our surgical technique is described in previous papers [5,6]. The mean bypass time was 140±60.8 min (range 25–485), and the mean duration of aortic cross-clamping was 70±33.6 min (range 10–227). There was a mean of 2.4±0.8 grafts (range 1–5), and a mean of 3.2±1.2 (range 1–7) distal anastomoses. Of the used grafts 1.0±0.6 (range 1–2) were arterial grafts with 1.2±0.9 (range 1–5) distal anastomoses and 428 patients (79.1 %) received at least one new arterial graft.

2.3. Statistical analysis
To test which variables can be considered risk factors for early mortality, Fisher's exact test (unifactor analysis) was used. Multiple logistic regression analysis was used to identify risk factors that independently contributed to increased or decreased risk. The odds ratios derived from the parameter estimates in the logistic regression analysis can be considered estimates of relative mortality risk. To identify a subset of variables with prognostic value for early mortality a stepwise selective procedure was used at a significance level p<=0.05 (significant level for entry, respectively, stay into the prognostic model). A receiver operating characteristics curve (ROC) was calculated to measure the prognositc value of this subset. This subset was then used to calculate a prognostic score ‘S’ and a predicted probability ‘P’ for early mortality. The prognostic score ‘S’ is a linear function of the variables included in the selected subset. If the variables are selected the ‘S’-score is represented by S=b0+b1X1+b2x2+···bhxh. The predicted probability (P) for early mortality is calculated by P=1/1+e-s. Sensitivity analysis, (Table 8) was used for evaluating the effect of the initial estimate on the final decision.


View this table:
[in this window]
[in a new window]
  		Table 8. 2x2 table for the evaluation of the early mortality predictiona

 

   3. Results
Top
 Abstract
 1. Introduction
 2. Materials and methods
 3. Results
4. Discussion
 Appendix A. Example of...
 References
 
The a priori (average) risk for early mortality is 75/541 (13.9%). Myocardial infarction was the direct cause of death in 25 patients, and in five patients the cause of death was not clear (Table 2). Results of the unifactor risk analysis are presented in Table 1. Risk factors for early mortality are: diabetes (p=10-4), hypertension (p=0.03), vascular disease (p=10-5), renal disease (p=0.02), lung disease (p=3x10-5), pre-myocardial infarction (p=0.003), between-myocardial infarction (p=2x10-5), no-sinus rhythm (p=0.03), no patent IMA graft (p=0.003) and no-elective operation (p=8x10-12). If the latter risk factor is present then the risk of early mortality amount to 48%. Multifactor risk analysis (Table 3) identified the operative period, diabetes, lung disease, between-myocardial infarction, acute- and emergency operations, as independent risk factors for early mortality. The odds ratios (estimates of relative risk) are, respectively: 0.31, 2.5, 1.9, 2.6, 2.0 and 10.9.


View this table:
[in this window]
[in a new window]
  		Table 2. Causes of 75 early deaths

 

View this table:
[in this window]
[in a new window]
  		Table 3. Multifactor risk analysis, logistic regression analysis, for early mortality (N=75 (13.9))

 
Using stepwise logistic regression analysis, the following variables were selected for prediction of early mortality: operative period, diabetes, vascular-, lung-disease, between myocardial infarction, acute- and emergency operations. The associated regression coefficients (bi), odds ratios, and p-values are presented in Table 4. The ROC curve gave an area under the curve value of 0.80.


View this table:
[in this window]
[in a new window]
  		Table 4. Stepwise logistic regression analysis (variables selected for prediction of early mortality)a

 
The S-score for an individual patient is calculated as follows:


The distribution of the S-scores and predicted probabilities P for early mortality in the group with (n=75) and without (n=466) early mortality is presented in Tables 5 and 6. The Scores are classified into the following classes: -4: (-4.5)–(-3.5); -3: (-3.5)–(-2.50)...2: (1.5)–(2.5). For the probability, the following classification is used: 0: 0<=P<0.10; 1: 0.10<=P<0.20;...0.80<=P<0.90. The observed probabilities in these discrete classes compare well with the mid-points of the predicted probabilities (Table 6) (note the number of patients in the classes 4–8 are very low).


View this table:
[in this window]
[in a new window]
  		Table 5. Distribution of the S-scores in group of patients with (n=75) and without (n=466) early mortality (S-score classified into discrete classes)a

 

View this table:
[in this window]
[in a new window]
  		Table 6. Distribution of predicted probabilities (P) in group of patients with (n=75) and without (n=466) early mortality. Classes defined as: 0=0<=P<0.10; 1=0.10<=P<0.20; ...8=0.80<=P<0.90

 
We decided to classify patients in low risk, intermediate risk, high risk and very high risk patients (Table 7). The a priori average risk of early mortality is 75/541 (14%). Using the S-score and the predicted probability P, we classified patients in low risk (5%), intermediate risk (15%), high risk (30%) and very high risk (40%). The observed mortality in these risk groups compare well with the predicted (4.9, 12, 31, 63%).


View this table:
[in this window]
[in a new window]
  		Table 7. Classification of patients in low risk, intermediate risk, high risk and very high risk patients

 
We used a predicted probability P>=0.40 (risk category very high) as a cut-off point for constructing a prognostic test for early mortality. The specificity of the test is high (0.97), but has a low sensitivity (0.33). The predictive values of a positive and a negative tests are 0.63 and 0.90, respectively (Table 8).


   4. Discussion
Top
 Abstract
 1. Introduction
 2. Materials and methods
 3. Results
 4. Discussion
Appendix A. Example of...
 References
 
The intention of this study is to construct a predictive test to evaluate preoperatively the risk for early mortality of a patient presented for RECABG. A better selection of patients undergoing RECABG is a first step in order to minimize mortality and morbidity. We focussed our study on early mortality because this 6-month-mortality is related to patient variables and is therefore an honourable endpoint to do a prediction of mortality risk of the operation [7]. In our previous study we analysed risk factors of early mortality in RECABG, and concluded that several pre-and perioperative variables were independent predictors of early mortality, with their influence on different moments (F.M.v.E., L.N., F.W.A.V., R.M.H.J.B. Analysis of mortality within the first six months after coronary reoperation. Ann Thorac Surg, accepted for publication). However, in the present study we included only preoperative variables in the analysis, because we want to evaluate and inform the patient preoperatively about his risk.

Cardiovascular causes were the reason for early mortality in 64 patients. In five patients we could not verify the cause of death, cardiac or not cardiac related; these patients were classified as unknown cause of mortality.

Using stepwise logistic regression analysis, the operative period, diabetes, vascular- and lung disease, between myocardial infarction, acute- and emergency operations were identified as predictive for early mortality. Independent of forward or backward performed regression analysis these variables were selected. These variables were comparable with variables selected in other studies [1,3,4]. The positive influence of the time period 1993–1998 can be related to several factors, growing surgical experiences in redosurgery, changing profile of the patients, as discussed in our previous report [6]. The area under the ROC curve (0.80) indicates that good discrimination remains between yes or no early mortality.

By using these seven variables we created a risk score. We classified patients in low risk (5%), intermediate risk (15%), high risk (30%) and very high risk (40%) for early mortality. The observed mortality in these risk groups compare well with the predicted (Table 7). The specificity of our test is high, as required for a prognostic test, but we have a low sensitivity. Still the test is useful to select patients with a very high risk of mortality. The predictive values of a positive and a negative test are 0.63 and 0.90, respectively (Table 8). This means that if the test is negative, there is only 10% risk of early mortality; on the other hand, if the test is positive, there is 63% risk that the patient will die during the first 6 months.

As for all other prediction models [8,9], it must be clear, that such a model never predict the specific outcome of an individual patient. Each patient will die or survive. However, with a predictive model we can inform ourselves and the patients about the probability of the risk of mortality.

However, even more than in primary myocardial revascularization, the outcome in RECABG, is also related to perioperative variables. The importance of operative status and perioperative myocardial infarction such as independent variables for mortality is documented in almost all studies concerning RECABG [1–6]. We included operative status as a preoperative variable in our predictive model, however, it is impossible to include a perioperative myocardial infarction as a preoperative variable in the predictive model. A perioperative myocardial infarction is related not only to several preoperative variables, such as diseased vein grafts, the native coronary system but also to a lot of operative variables. Route of delivery of cardioplegia, completeness of the revascualization, perioperative spreading of debris from vein grafts in distal coronary arteries, are operative variables related to the occurrence of a perioperative myocardial infarction. The impact of a perioperative myocardial infarction on the early mortality is, however, important [1–4]. We must therefore realise that our prediction of likely risk of early mortality after RECABG, flatters the reality if a patient has a perioperative myocardial infarction. Furthermore this is a single institution experience, and several variables are not incorporated in the analysis. Ejection fraction is not calculated for all patients, and therefore not incorporated in the analysis. However, by using the variables myocardial infarction and between infarction we try to have an idea about ventricular function. Other variables such as lack of graft material, a calcified aorta, previous mediastinitis are certainly variables which increases the operative risk.

On the other hand, with our predictive model, we have an evidence-based instrument to identify preoperatively high risk patients for RECABG. This can influence our strategy on several points, the decision to operate or not, operative strategy, type of anaesthesia and postoperative care.

In conclusion, with our predictive model we can stratify patients presented for RECABG, according to their risk for early mortality. Based on this stratification we can not only inform the patient concerning the risk, but also discus the opportunity of the RECABG. At this point we can improve our results by making a better preoperative patient selection.


   Acknowledgments
 
We thank Johannes M. van Druten, PhD, Department of Medical Informatics, Epidemiology and Statistics, University of Nijmegen, for the statistical analysis.


   Appendix A. Example of the predictive model
Top
 Abstract
 1. Introduction
 2. Materials and methods
 3. Results
 4. Discussion
 Appendix A. Example of...
References
 
A.1. Formula

Calculation for the S-score:



Calculation of the predicted probability:

A.2. Case

Patient:
A 73-year-old male, insulin dependent diabetic, history of vascular disease, no lung disease, history of a myocardial infarction after the first operation.
First calculation if this patient is presented for an elective operation
Second calculation if this patient is presented for an emergency operation

Elective operation:



Classification: intermediate risk category (0.10<=P<0.20), predicted probabilty 15%
Emergency operation:



Classification: high risk category (0.20<=P<0.40), predicted probalility 30%


   References
Top
 Abstract
 1. Introduction
 2. Materials and methods
 3. Results
 4. Discussion
 Appendix A. Example of...
 References
 

  1. He G.-W., Acuff T.E., He Y.-H., Ryan W.H., Mack M.J. Determinants of operative mortality in reoperative coronary bypass grafting. J Thorac Cardiovasc Surg 1995;110:971-978.[Abstract/Free Full Text]
  2. Yau T.M., Borger M.A., Weisel R.D., Ivanov J. The changing pattern of reoperative coronary surgery: trends in 1230 consecutive reoperations. J Thorac Cardiovasc Surg 2000;120:156-163.[Abstract/Free Full Text]
  3. Weintraub W.S., Jones E.L., Craver J.M., Grosswald R., Guyton R.A. In-hospital and long-term outcome after reoperative coronary artery bypass graft surgery. Circulation 1995;92(Suppl II):II-50-II-57.
  4. Christenson J.T., Schmuziger M., Simonet F. Reoperative coronary artery bypass procedures: risk factors for early mortality and late survivial. Eur J Cardiothorac Surg 1997;11:129-133.[Abstract]
  5. Noyez L., Skotnicki S.H., Lacquet L.K. Morbidity and mortality in 200 consecutive coronary reoperations. Eur J Cardiothorac Surg 1997;11:528-532.[Abstract]
  6. Eck van F.M., Noyez L., Verheugt F.W.A., Brouwer R.M.H.J. Changing profile of patients undergoing redo-coronary artery surgery. Eur J Cardiothorac Surg 2001;21:205-211.[Abstract/Free Full Text]
  7. Blackstone E.H. Outcome analysis using hazard function methodology. Ann Thorac Surg 1996;61(2 Suppl):S2-S7.
  8. Parsonnet V., Dean D., Bernstein A.D. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation 1989;79(Suppl I):I-3-I-12.
  9. Nashef S.A.M., Roques F., Michel P., Gauducheau E., Lemeshow S., Salamon R. European system for cardiac operative risk evaluation (EUROscore). Eur J Cardiothorac Surg 1999;16:9-13.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Ann. Thorac. Surg.Home page
J. F. Sabik III, E. H. Blackstone, P. L. Houghtaling, P. A. Walts, and B. W. Lytle
Is Reoperation Still a Risk Factor in Coronary Artery Bypass Surgery?
Ann. Thorac. Surg., November 1, 2005; 80(5): 1719 - 1727.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Thorac. Surg.Home page
N. Luciani, A. Anselmi, M. Gaudino, G. Nasso, F. Glieca, L. Martinelli, F. Santarelli, M. Perisano, and G. Possati
Harmonic Scalpel Reduces Bleeding and Postoperative Complications in Redo Cardiac Surgery
Ann. Thorac. Surg., September 1, 2005; 80(3): 934 - 938.
[Abstract] [Full Text] [PDF]

Home page
 Eur. J. Cardiothorac. Surg.Home page
L. Noyez and F. M. van Eck
Long-term cardiac survival after reoperative coronary artery bypass grafting
Eur. J. Cardiothorac. Surg., January 1, 2004; 25(1): 59 - 64.
[Abstract] [Full Text] [PDF]

Home page
 Ann. Thorac. Surg.Home page
F. M. van Eck, L. Noyez, F. W.A. Verheugt, and R. M.H.J. Brouwer
Analysis of mortality within the first six months after coronary reoperation
Ann. Thorac. Surg., December 1, 2002; 74(6): 2106 - 2112.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to Personal Folders
Right arrow Download to citation manager
Right arrow Author home page(s):
Luc Noyez
Rene M.H.J. Brouwer
Right arrow Permission Requests
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by van Eck, F. M.
Right arrow Articles by Brouwer, R. M.H.J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by van Eck, F. M.
Right arrow Articles by Brouwer, R. M.H.J.
Related Collections
Right arrow Coronary disease

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
ANN THORAC SURG ASIAN CARDIOVASC THORAC ANN EUR J CARDIOTHORAC SURG
J THORAC CARDIOVASC SURG ICVTS ALL CTSNet JOURNALS