Ann Thorac Surg 2004;78:2022-2026
© 2004 The Society of Thoracic Surgeons
Original article: cardiovascular
Development of a Score to Predict the Need for Coronary Artery Bypass Graft Surgery in Patients With Non-ST Segment Elevation Acute Coronary Syndromes
Santiago Garcia, MDa,*,
Mariana J. Canoniero, MDa,
Julio A. Chirinos, MDa,
Eduardo de Marchena, MDa,
Tomas Salerno, MDb,
Alexandre Ferreira, MDa
a Division of Cardiology, Department of Medicine, and Division of Cardiothoracic Surgery, Department of Surgery, Jackson Memorial Hospital, University of Miami School of Medicine, Miami, Florida, USA
b Division of Cardiothoracic Surgery, Department of Surgery, Jackson Memorial Hospital, University of Miami School of Medicine, Miami, Florida USA
Accepted for publication March 8, 2004.
* Address reprint requests to Dr Garcia, 8395 NW 143rd Terrace, Miami, FL 33016, USA
agarcia7{at}med.miami.edu
 |
Abstract
|
|---|
BACKGROUND: Our ability to identify surgical candidates before angiography is limited. Early identification of surgical patients would improve preoperative management and ultimately postoperative outcomes. The objective of this study was to determine whether surgical candidates could be identified before coronary angiography using simple clinical variables.
METHODS: The study population was comprised of 688 patients admitted to a tertiary hospital because of non-ST segment elevation acute coronary syndromes. Stepwise logistic regression analysis was performed to identify predictors of surgery. A test cohort (50.2%) was used to generate the model and a validation cohort (49.8%) was used for independent validation of the proposed score.
RESULTS: Three variables independently predicted the indication for bypass surgery: the absolute thrombolysis in myocardial infarction (TIMI) risk score (odds ratio [OR] = 2.34 for each unit increase in the score, 95% confidence interval [CI] = 1.89–2.89, p < 0.001), the presence of peripheral vascular disease (PVD) (OR = 4.08, CI = 1.48–11.24, p = 0.006), and the presence of congestive heart failure (CHF) on admission (OR = 2.57, CI = 1.08–6.81, p = 0.03). A simplified score that spans from 0–10 was developed based on the logistic regression model. The score adds two points to the TIMI score if PVD is present and one point if CHF is present. The area under the receiver-operating-characteristic (ROC) curve of the proposed score for predicting surgery was 0.80 ± 0.02.
CONCLUSIONS: The score we have proposed and validated can be used to predict the likelihood of bypass surgery before coronary angiography and may assist the clinician to tailor preoperative medical therapy.
|
Introduction
|
|---|
Approximately 20%–30% of the more than 1 million patients hospitalized because of non-ST elevation acute coronary syndromes (NSTEACS) in the United States will require surgical revascularization or coronary artery bypass grafting (CABG) [1]. Although older age, the presence of diabetes mellitus (DM), and left ventricular dysfunction have been associated with a greater probability of CABG [2, 3], our ability to identify surgical candidates before angiography is limited [4–9]. With the increasing usage of antiplatelet and antithrombotic therapies during the preoperative period, untoward events in the postoperative period are at risk of developing in surgical patients [10–12]. A predictive model designed to identify surgical patients using preangiographic variables assessed at the time of admission would allow the clinician to individualize preoperative medical therapy based on the likelihood of CABG, which may result in improved outcomes. Our objective regarding this study was to develop a score that is easily calculated at the time of patient presentation that discriminates between surgical and nonsurgical patients before coronary angiography.
|
Material and Methods
|
|---|
Patient Population and Data Collection
The cardiac catheterization database at a large volume, university-affiliated, tertiary medical center was reviewed for data collection. Patients who underwent early coronary angiography (within 7 days of presentation) from January 1, 2000 to June 30, 2002 were identified. Patients (n = 1117) who underwent cardiac catheterization secondary to new onset or worsening chest pain, occurring at rest or with minimal exertion, and alleviated by nitroglycerin and/or rest were considered eligible for the study. Two independent investigators reviewed the medical records including electrocardiograms and coronary angiography. Patients who exhibited ST-segment elevation on their admission electrocardiogram indicative of acute transmural myocardial infarction (MI), new onset left bundle branch block, previous CABG, a percutaneous coronary intervention (PCI) within the previous 6 months, or a definitive nonischemic etiology for their chest pain were excluded. The study was approved by the Institutional Review Board at the University of Miami School of Medicine (March 2003) in accordance with Title 45 of the Code of Federal Regulations (Part 46: Protection of Human Subjects) policy.
Statistical Analysis
Continuous variables demonstrating a normal distribution were expressed as mean ± standard deviation (SD) and discrete variables were presented as frequencies and percentages. Continuous variables were compared between the two groups using the nonpaired Student's t test for normally distributed data and the Mann–Whitney U test for nonnormally distributed data. Proportions were compared with the 
2 test. Statistical significance was defined as a two-tailed p value less than 0.05. Univariate analysis was performed to identify factors significantly associated with an increased risk of requiring CABG. The following variables were included in the univariate analysis: age greater than 65 years, diabetes mellitus (DM), hypertension (HTN), dyslipidemia, current smoking history, previous PCI, stroke, congestive heart failure (CHF), peripheral vascular disease (PVD), and previous coronary artery disease (CAD) (following the example of the thrombolysis in myocardial infarction (TIMI) Group [13], we also broadened the angiographic definition of known CAD to include patients with a history of Q-wave MI or previous coronary revascularization), severe anginal symptoms (
2 episodes of chest pain at rest in the last 24 hours), evidence of heart failure (one of the following: orthopnea, paroxysmal nocturnal dyspnea, S3, crackles, distension of neck veins, peripheral edema, or pulmonary congestion on roentgenogram), ST-segment depression greater than or equal to 0.5 mm, and elevated troponin I (> 1 ng/ml).
Variables were described according to the American College of Cardiology Task Force on clinical data standards (Acute Coronary Syndromes Writing Committee) [14]. Additionally, we included all of the components of the TIMI risk score [15] in the analysis. The TIMI risk score and its components are presented in Table 1.
View this table:
[in this window]
[in a new window]
|
Table 1. Thrombolysis in Myocardial Infarction (TIMI) Risk Score for Unstable Angina/Non-ST Segment Elevation Myocardial Infarction
|
|
The primary outcome was the indication for CABG after angiography was performed as recommended by the interventional cardiology team. Those variables that demonstrated significant correlation with the outcome (p < 0.05) underwent multivariate analysis.
Multivariate Analysis
Those variables that reflected a significant correlation with the outcome in univariate analysis underwent multivariate analysis to identify independent predictors of bypass surgery. The population was randomly divided into a test cohort (50.2% of patients) and a validation cohort (49.8% of patients). The test cohort was used to generate the model and the validation cohort was used for independent validation of the model output. Stepwise logistic regression (forward selection) was used to construct predictive models. Only variables that achieved a significance level of 
less than 0.05 were eligible to enter the stepwise multiple logistic analysis and be retained in the final model.
A second stepwise logistic regression model was obtained after incorporating the TIMI risk score into the model and eliminating those variables that were previously included in the score. The discriminative ability of the models was assessed by calculating the area under the receiver-operating-characteristic (ROC) curve. NCSS 2001 for Windows (NCSS, Kaysville, UT) and Medcalc for Windows version 7 (Medcalc Software, Mariakerke, Belgium) were used to perform the statistical analysis.
|
Results
|
|---|
Of the 2624 patients who underwent coronary angiography within the study period, 1117 (42%) presented to the emergency department with chest pain consistent with acute coronary syndrome. Of these, 688 patients (62%) without ST-segment elevation, history of CABG, or PCI in the previous 6 months underwent cardiac catheterization within 7 days of admission and comprise the study population.
CABG was recommended for 249 patients (36%). Univariate and multivariate predictors of CABG are presented in Table 2.
Stepwise logistic regression was conducted including those variables that indicated a significant correlation with the outcome in univariate analysis (Table 2). An outcome prediction model constructed with multivariate predictors of CABG was assessed in the test and validation cohort. This model correctly classified 73% of patients in the test cohort and 71% of patients in the validation cohort as to whether surgery was or was not indicated. The area under the ROC curve with regard to discriminating between surgical and nonsurgical patients was 0.76 ± 0.02.
A second stepwise logistic regression model was obtained that incorporated the TIMI risk score as a single variable and eliminated those variables that were already included in the score. Given that many components of the TIMI risk score were among the multivariate predictors of surgery, we decided to incorporate the summary score itself and aimed to reduce the complexity of the model while maintaining its diagnostic accuracy. The TIMI risk score was preferred over other summary scores because of its widespread use among cardiologists and clinicians and because it correlates with multivessel and left main disease [16], which are surrogate markers for CABG.
After including the TIMI risk score, three variables remained in the model (Table 3). This model correctly classified 76.5% of the patients in the test cohort and 75% of the patients in the validation cohort with an area under the ROC curve of 0.80 ± 0.2.
View this table:
[in this window]
[in a new window]
|
Table 3. Stepwise Logistic Regression Model Obtained After Incorporating the Thrombolysis in Myocardial Infarction (TIMI) Risk Score and Excluding Its Individual Components
|
|
A simplified score was created by transforming the TIMI score coefficient to 1 and rounding the remaining estimated logistic regression coefficients to the nearest proportional integer as follows: TIMI score number [0–7] + 2 if PVD is present + 1 if CHF is present. This score, which spans from 0–10, stratified patients over a wide range of risk (Fig 1). The ROC curve, sensitivities, specificities, and predictive values of different score cutoff levels with regard to discriminating between surgical and nonsurgical patients are depicted in Figure 2.
View larger version (35K):
[in this window]
[in a new window]
|
Fig 2. Receiver-operating-characteristic (ROC) curve for various cutoff levels of the proposed score in differentiating between surgical and nonsurgical patients. The cutoff level with maximum diagnostic accuracy is in bold. (CI = confidence interval; NPV = negative predictive value; PPV = positive predictive value.)
|
|
|
Comment
|
|---|
We have developed and validated a simple score to predict the likelihood of performing CABG in patients admitted with NSTEACS. This score, which can be calculated at the bedside without the assistance of computers, encompasses three elements assessed at the time of admission: the TIMI risk score, the presence of PVD, and the presence of CHF.
Using a cutoff of 4, the score manifests favorable specificity (92.71%) to predict surgical disease with a sensitivity of 46.78%. Assuming a prevalence of surgical disease similar to that encountered in this population, the following estimation can be made: for each 100 patients with NSTEACS, 35 of them will receive surgery and 65 will not. Among the 35 patients who will ultimately require CABG, the score will correctly classify 16 patients as having surgical disease and will misclassify the remaining 19 patients. In the remaining 65 patients who will receive PCI or medical treatment alone, the score will correctly identify 60 patients (92%), whereas 5 will be misclassified as surgical patients. Overall 76 out the 100 patients will be correctly classified using this cutoff.
The use of the score to decide whether or not clopidogrel should be administered is a good example of its clinical application. Clopidogrel has raised genuine safety concerns among cardiovascular surgeons and cardiologists alike given that, if administered preoperatively (< 5 days), transfusion requirements increase by 40%, the risk of major bleeding increases by 3.3%, and the risk of surgical reexploration for the management of postoperative bleeding increases tenfold [10–12].
If the score was used in a population with similar characteristics to the study group, the following estimations can be made: for each 100 patients, 16 surgical patients will not receive clopidogrel resulting in an appropriate intervention (no surgical delay), whereas 19 surgical patients will receive clopidogrel thereby delaying surgical intervention. Among nonsurgical patients, 60 out of 65 patients will receive clopidogrel resulting in an appropriate intervention, whereas 5 nonsurgical patients will not benefit from clopidogrel because of misclassification. In summary if the proposed score is used, 76 out of 100 patients would be correctly classified and will either benefit from early clopidogrel administration or not suffer delays regarding the performance of a CABG procedure if required. Therefore this score would not only allow the clinician to expand the indication of a drug with proven benefit but would further reduce the number of patients in whom a CABG procedure is postponed by 50% because of premature clopidogrel administration. Although we present these calculations for the optimal score cutoff in a number of patients, it should be emphasized that the score has a range of values that overlaps with regard to patients who require surgery and those who do not. The risk increases progressively with increasing scores (Fig 1). There is no threshold at which the risk suddenly rises; therefore various cutoff points can be used. Very high scores are highly specific, whereas low scores are highly sensitive for detecting surgical patients.
In addition by using the risk score patients could be advised before angiography of the likelihood of undergoing a surgical revascularization, which may be urgent after angiography. Furthermore the score may allow for greater precision with regard to the timing of the angiography based on the necessity for a cardiovascular surgeon and available operating room.
Finally the proposed score could be a useful tool with regard to selecting high-risk patients for clinical studies, evaluating new treatment modalities, and figuring out how to minimize time to revascularization and decrease the rate of complications of NSTEACS patients awaiting surgical treatment.
Study Limitations
This study is limited by the biases innate to a retrospective registry. Data collection may not be as accurate as data collected prospectively. Also the decision on which revascularization strategy would be most appropriate in each case was decided at the discretion of the attending cardiologist and surgeon. However the main limitation of this study seems to be the external validity. Recent studies have indicated that 60% of patients with multivessel disease can be targeted with either angioplasty or bypass surgery [17]. This study, though, reflects the approach of a tertiary, university-based, medical center located within the United States and this may not be the standard of care at other institutions. These results apply to the population of patients referred for angiography from this dataset, but may not be representative of the entire population of patients with NSTEACS. The fact that 36% of the patients in this study were treated surgically may be secondary to the exclusion of patients with previous CABG in whom a second intervention is often avoided, especially if the graft that supplies the LAD is patent [2]. The use of this score to predict absolute risk in new populations therefore requires further evaluation given the potential differences inherent within population characteristics.
Conclusion
A score that includes the TIMI risk score, the presence of PVD, and the presence of CHF can be used to predict the likelihood of bypass surgery before coronary angiography and may assist the clinician to tailor preoperative medical therapy.
|
Acknowledgments
|
|---|
The authors are indebted to Dr Elana Oberstein for her thorough review of the manuscript.
|
References
|
|---|
- National Center for Health Statistics. Detailed Diagnoses and Procedures, National Hospital Discharge Survey, 1996. Hyattsville: National Center for Health Statistics, 1998, p 13 (data from Vital and Health Statistics)
- Braunwald E, Antman EM, Beasley JW, et al. ACC/AHA 2002 guideline update for the management of patients with unstable angina and non-ST-segment elevation myocardial infarction. A report of the American College of Cardiology/American Heart Association Task Force on practice guidelines (Committee on the Management of Patients with Unstable Angina). J Am Coll Cardiol. 2000;36:970–1062[Free Full Text]
- Braunwald E, Antman EM, Beasley JW, et al. ACC/AHA 2002 Guideline Update for the Management of Patients With Unstable Angina and Non-ST-Segment Elevation MI. A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. (Committee on the Management of Patients With Unstable Angina). Available at http://www.acc.org/clinical/guidelines/unstable/unstable.pdf. Accessed February 15, 2004
- de Zwaan C, Gorgels A, Wellens H, et al. Unstable Angina. Are we able to recognize high-risk patients? Chest. 1997;112:244–250[Abstract/Free Full Text]
- Dangas G, Mehran R, Wallenstein S, et al. Correlation of angiographic morphology and clinical presentation in unstable angina. J Am Coll Cardiol. 1997;29:519–525[Abstract]
- Plotnick G, Fisher M, Becker L. Ventricular arrhythmias in patients with rest angina: correlation with ST-segment changes and extent of coronary atherosclerosis. Am Heart J. 1983;105:32–36[Medline]
- Ahmed W, Bittl J, Braunwald E. Relation between clinical presentation and angiographic findings in unstable angina pectoris, and comparison with that in stable angina. Am J Cardiol. 1993;72:544–550[Medline]
- Hubbard B, Gibbons R, Lapeyre A, et al. Identification of severe coronary artery disease using simple clinical parameters. Arch Intern Med. 1992;152:309–312[Abstract/Free Full Text]
- Pirwitz M, Lange R, Landau C, et al. Utility of the 12-lead electrocardiogram in identifying underlying coronary artery disease in patients with depressed left ventricular systolic function. Am J Cardiol. 1996;77:1289–1292[Medline]
- The clopidogrel in unstable angina to prevent recurrent events trial investigators. Effects of clopidogrel in addition to aspirin in patients with acute coronary syndromes without ST-segment elevation. N Engl J Med. 2001;345:494–502[Abstract/Free Full Text]
- Khot UN, Nissen SE. Is CURE a cure for acute coronary syndromes? Statistical versus clinical significance. J Am Coll Cardiol. 2002;40:218–219[Abstract/Free Full Text]
- Hongo R, Ley J, Dick S, et al. The effect of clopidogrel in combination with aspirin when given before coronary artery bypass grafting. J Am Coll Cardiol. 2002;40:231–237[Abstract/Free Full Text]
- Morrow DA, Antman EM, Snapinn SM, et al. An integrated clinical approach to predicting the benefit of tirofiban in non-ST elevation acute coronary syndromes. application of the TIMI Risk Score for UA/NSTEMI in PRISM-PLUS. Eur Heart J. 2002;23:223–229[Abstract/Free Full Text]
- Cannon CP, Battler A, Brindis RG, et al. ACC key elements and data definitions for measuring the clinical management and outcomes of patients with acute coronary syndromes. A report of the American College of Cardiology Task Force on clinical data standards (Acute Coronary Syndromes Writing Committee). J Am Coll Cardiol. 2001;38:2114–2130[Free Full Text]
- Antman E, Cohen M, Bernick P, et al. The TIMI risk score for unstable/non-ST elevation MI. A method for prognostication and therapeutic decision making. JAMA. 2000;284:835–842[Abstract/Free Full Text]
- Garcia S, Canoniero M, Peter A. Correlation of TIMI risk score with angiographic severity and extent of coronary artery disease in patients with non-ST-segment elevation acute coronary syndromes. Am J Cardiol. 2004;93:813–816[Medline]
- Rigter H, Meijler AP, Mc Donnell J, et al. Indications for coronary revascularization: a Dutch perspective. Heart. 1997;77:211–218[Abstract/Free Full Text]
Related Article
-
INVITED COMMENTARY
- Verdi J. DiSesa
Ann. Thorac. Surg. 2004 78: 2026-2027.
[Extract]
[Full Text]
[PDF]
Top
Abstract
Introduction
