| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 2000;120:1112-1119
© 2000 The American Association for Thoracic Surgery
Surgery for Acquired Cardiovascular Disease |
From the Departments of Surgery,a Medicine,b and Anesthesiology,c University of Alabama at Birmingham, Birmingham; Department of Medicine,d Duke University Medical Center, Durham, NC; Alabama Quality Assurance Foundation,e Birmingham; and Birmingham Veterans Affairs Medical Center,f Birmingham, Ala.
Address for reprints: William L. Holman, MD, Department of Surgery, 703 South 19th St, University of Alabama at Birmingham, Birmingham, AL 35294-0007 (E-mail: wholman{at}its.uab.edu).
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
1,2 and prospective 3,4 single center studies that show decreased morbidity and mortality with the use of preincision intra-aortic balloon pumps (IABPs) in high-risk patients undergoing coronary artery bypass grafting (CABG). However, these studies included many patients with ongoing myocardial infarction and preoperative cardiogenic shock, which are generally accepted therapeutic rather than prophylactic indications for IABPs.
The Alabama CABG Cooperative Project is a statewide process-oriented project involving all hospitals performing CABG in the state. 5 As part of the project, numerous demographic, procedural, and outcome variables were gathered. Two of these variables were use of an IABP (yes/no) and time of IABP insertion relative to the CABG procedure (preincision/intraoperative/postoperative).
The present study analyzes data gathered by the Alabama CABG Cooperative Project to determine the effect of prophylactic preincision IABP use in hemodynamically stable, but high-risk patients undergoing CABG by comparing risk-adjusted patient outcomes. Patients with evidence of very recent or impending myocardial damage or with hemodynamic instability were not included in this analysis. Only the demographic variables and descriptors of patient disease that were available before the operation were used to define risk. The hypothesis tested is that prophylactic preincision IABP placement in hemodynamically stable, high-risk patients undergoing CABG favorably influences in-hospital mortality and post-CABG length of hospital stay (LOS).
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
5 For the initial phase of the Alabama CABG Cooperative project, a sample of Medicare patients who underwent CABG surgery was selected. This sample consisted of all Medicare patients discharged from participating hospitals between July 1, 1995, and June 30, 1996 (initial round of data abstraction) and between January 1, 1998, and June 30, 1998 (second round of data abstraction). Patients with a diagnosis-related group (DRG) of 106 (coronary artery catheterization and CABG surgery during the same admission) or DRG 107 (only CABG surgery during the admission) were included. Patients with DRG 104 (cardiac valve procedure with catheterization), DRG 105 (cardiac valve procedure without catheterization), and DRG 468 (extensive procedure unrelated to principal diagnosis) were excluded. Initially we had data on 9235 CABG procedures. After deleting missing data on clinical outcome, sex, and race, records for 8972 hospitalizations were left for analysis. In our analysis, placement of an IABP is considered therapeutic rather than prophylactic for patients with impending or very recent myocardial damage or for patients with hemodynamic instability. We therefore excluded the following patients: (1) those in shock and/or undergoing mechanical ventilation before the operation (N = 377) and (2) those undergoing emergency CABG, those who had undergone percutaneous transluminal coronary angioplasty (PTCA) within the 6 hours preceding CABG, or those who had a documented myocardial infarction no more than 3 days before the operation (N = 1198). There were 184 patients in the groups 1 and 2 described in the previous sentence. Hence, we were left with 7581 patients for our analyses, of whom 592 received a preincision IABP. Patients undergoing intraoperative or postoperative IABP were included in the group not receiving preincision IABP.
Analyses
After performing descriptive statistics on the CABG Project dataset, we engaged in the following steps.
Step l
Generation of a propensity score to model the likelihood of prophylactic preincision IABP use, based on patient characteristics available to the surgeon preoperatively, and likely to influence the surgeon's decision to place an IABP.
Step 2
Bivariate analyses of associations between patient characteristics, propensity scores, receipt of preincision IABP, and study outcomes: mortality in-hospital and at 30, 60, 180, 360, and 1000 days, as well as postoperative LOS.
Step 3
Development of multivariable models for each of these outcomes (mortality and LOS) as dependent variable, and with receipt of preincision IABP as the main independent variable, and propensity score as well as other clinical factors as covariates.
Step 4
Matching of each patient receiving preincision IABP with another patient of similar propensity score and similar presentation based on other key variables. Outcomes for these matched pairs were compared.
We therefore tested our hypothesis by using 2 approaches. First, in above steps 2 and 3, we examined the association between receipt of preincision IABP and outcome, before and after adjusting for the a priori likelihood that the patient would receive a preincision IABP, based on his/her likelihood to receive one (propensity score), and on other clinical characteristics. Second, in step 4, we directly compared a set of patients who received a prophylactic preincision IABP with another who did not, even though for each pair their likelihood of receiving a preincision IABP was similar.
Development of the propensity score model
The propensity score method, described by Rosenbaum and Rubin, 6,7 allows a direct estimation of the likelihood of receipt of the treatment under study (in our case, preincision IABP) on the basis of the variables deemed important in making the treatment decision. The propensity score can then be used to adjust globally for the characteristics that were deemed influential in the decision to treat or not to treat and becomes a powerful tool for adjusting for treatment selection bias (confounding by indication) in observational studies such as ours. In this study, the propensity score is the probability of IABP insertion predicted by a multiple logistic regression (MLR) model where actual preincision IABP receipt is the dependent variable and the variables mentioned next are the independent variables.
Currently, cardiac surgeons place or do not place a preincision IABP according to their subjective assessment of its potential benefits. We queried 2 cardiac surgeons who reached consensus on which of the variables collected by the Alabama CABG Cooperative Project were influential in the decision to insert a preincision IABP. Those variables chosen include patient age, sex, comorbidity (index adapted from Charlson), history of diabetes mellitus, history of prior CABG, renal insufficiency at presentation (creatinine greater than 2.0 mg/dL or blood urea nitrogen of 35 mg/dL or greater, or on dialysis), congestive heart failure at presentation (ejection fraction of 0.30 or less by imaging procedure, clinical presentation consistent with heart failure, or history of cardiac decompensation within 2 weeks before the operation), or left main coronary artery stenosis of 70% or greater. These were the variables used to develop our propensity score. Other patient characteristics were deemed indications for a preincision IABP, but as treatment rather than prophylaxis. Hence, these variables were used as exclusion criteria and not in the propensity score or elsewhere in our analyses. These variables included the following: hemodynamic instability (shock as determined by blood pressure after arrival at hospital and before incision), initiation of mechanical ventilation before entering the operating room, PTCA performed at most 6 hours before CABG, and very recent myocardial infarction (at most 3 days before CABG).
We developed our propensity score model in several iterative steps in which we categorized our variables in a way to optimize the fit of the model to the data. Because we were focusing on optimal prediction, we left every initial variable in the final model, regardless of the level of statistical significance of its coefficient. We tested whether the propensity score would adjust completely for all the covariates it incorporates by checking for differences in individual covariates for patients with and without an IABP after stratifying for quintiles of propensity score. Because several variables showed statistically significant differences in some of the quintiles of propensity scores, we also used them as covariates in the subsequent MLR modeling outcomes. In addition, we made similarity of those variables a matching condition when we created matched pairs of patients receiving and not receiving a prophylactic preincision IABP.
Case-matching procedure
Each patient who received a preincision IABP was matched with another patient who did not receive a preincision IABP but with a similar propensity score (the difference in propensity score for members of the same pair had to be < 0.03). We also matched the pairs on those covariates where the analysis by quintile of propensity score had suggested that there were differences between those who did versus those who did not receive a preincision IABP. Thus, we ensured that the members of each matched pair were similar in terms of those characteristics deemed influential in the decision regarding a prophylactic preincision IABP.
Mortality modeling
For each patient, we had vital status as of June 22, 1999, ascertained from Medicare denominator files. Therefore, we were able to determine in-hospital mortality for all patients, as well as 30-, 60-, 180-, 360- and 1000-day mortality for most. In fact, we had length of survival, with censoring at 374 days for all. Therefore, we were able to perform MLR modeling to predict mortality by using for dichotomous outcomes death in-hospital, or death at 30, 60, 180, 360, and 1000 days. We did this by using the propensity score and other clinical characteristics as predictor variables.
We performed Cox proportional hazards analyses with death as the outcome variable and censoring as of June 22, 1999 (hence variable follow-up time for each patient). For all these models, actual receipt of preincision IABP was the main independent variable, whereas propensity score and other clinical variables were covariates. These covariates included the variables for which quintile of propensity score analysis had shown that the propensity score might not fully adjust, as well as variables to adjust for hospital characteristics, such as bed size (dichotomized at 400 beds). We developed a statistical model that predicted a patient's expected risk for operative mortality according to preoperative clinical factors. Potential clinical risk factors were chosen on the basis of a number of previously published CABG risk models. 8-10 The details of this risk adjustment model have been published. 5 For this study, we also built models with the study outcomes as dependent variables, receipt of preincision IABP as main independent variable, and the previously derived risk-adjusted mortality as a covariate. The coefficients for preincision IABP were similar to those obtained when we used propensity scores, so we do not present these other models.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
|
|
|
|
|
When we compared unadjusted survivals in the entire population, as opposed to just the matched pairs, patients who had received preincision IABPs had consistently lower survivals than those who had not. When we refined these comparisons by multivariable adjustment via Cox proportional hazards modeling on the entire population, whether preincision IABP had been received was no longer significantly associated with survival (Table V). In fact, the hazard ratio for IABP was less than 1 (0.90), which would suggest some beneficial effect for IABP, except that the 0.90 hazard ratio had a 95% confidence interval of 0.72 to 1.13, which overlaps 1. Hence, no inference of a beneficial effect should be made.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
To measure the benefit of prophylactic preincision IABPs, we developed propensity scores for individual patients. By using this propensity score and also adjusting for additional risk factors for post-CABG death, our study failed to convincingly demonstrate a lower adjusted mortality for patients who received a prophylactic preincision IABP, as compared with risk-matched patients who did not receive a preincision IABP. Thus, there did not appear to be any survival advantage in placing an IABP before the operation in these high-risk patients, as opposed to placing an IABP on an "as needed" basis during or after the operation. We did, however, find that prophylactic IABP use was associated with a significantly shorter postoperative LOS, indicating that these patients had an improved convalescence. Furthermore, the trend in survival that is evident at 1 to 3 years after the operation (Fig 1
) may become a significant survival benefit for prophylactic IABPs as the duration of follow-up increases.
The avoidance of morbid events by prophylactic use of an IABP, rather than by placing an IABP after acute myocardial ischemia has already occurred, represents an important change in the care of patients undergoing CABG. It is, however, one that should not be taken lightly, because the prediction of morbid events or death in individual patients is still somewhat imprecise. In other words, the benefit of prophylactic IABP use in hemodynamically stable patients is achieved at the cost of placing IABPs in some patients who would have tolerated CABG without IABP support.
Our findings in this study agree with Christenson's observation that a preincision IABP improves the high-risk patient's postoperative course, 3,4 even though we could not show a survival benefit. Our survival data are also different from the retrospective single center experiences of Dietl and colleagues 1 and Gutfinger and coworkers. 2 The apparent inconsistencies highlight the difficulty in addressing this complex issue and warrant a closer look at the other studies 1-4 that demonstrated lower mortality with preincision IABP use.
The prospective randomized trials published by Christenson and colleagues 3,4 are the most compelling evidence in support of preincision IABP use for high-risk patients. However, 50% to 60% of the patients in each of Christenson's studies had unstable angina or ongoing infarction, which is an important difference from our study that excluded these patients from the prophylactic IABP group. It is likely that preincision IABP use favorably affects short-term post-CABG mortality when it is placed before the operation as treatment for severe unstable angina or ongoing infarction. This statement is based on published experience with IABP support, 2,11 the ability of a prophylactic IABP to ameliorate acute intraoperative ischemia (a cause of perioperative myocardial infarction 12), and the difficulty in promptly diagnosing new or worsening myocardial ischemia during CABG procedures. 13-15
The retrospective analysis of Dietl and colleagues 1 examined 30-day mortality for patients with a left ventricular ejection fraction of 0.25 or less who had CABG with or without preoperative insertion of an IABP. Despite a significantly higher prevalence of previous CABG, New York Heart Association class III or IV symptoms, emergency surgery, recent myocardial infarction, and left main stenosis in patients who have had IABP support, there was a significantly lower mortality as compared with non-IABP patients. Of note, the patients in Dietl's series who received a preincision IABP had a significantly higher prevalence of urgent or emergency operation, intravenous nitroglycerin use, recent (0-7 days) myocardial infarction, and unstable angina. The issue of patients who are hemodynamically stable and who do not have the more severe types of unstable angina or myocardial infarction was not addressed. It is this group of patients who pose the greatest challenge for decisions regarding prophylactic use of an IABP and who were the subjects in our analysis.
The article from Gutfinger and associates 2 is another retrospective study of preoperative IABP use in high-risk patients undergoing CABG. High risk in this study was defined by age of 70 years or older, urgent operation for failed PTCA, emergency CABG reoperation, left main stenosis of 70% or more, left ventricular ejection fraction of 0.40 or less, and angina refractory to medical management. The patients who received an IABP were compared with patients who were 70 years or older but did not receive a preoperative IABP. The preoperative IABP patients had a significantly higher prevalence of acute myocardial infarction and congestive heart failure before the operation. This group also had significantly lower left ventricular ejection fractions and higher Parsonnet scores. The absence of a statistically significant difference in mortality between groups was considered to be evidence for a benefit. The outcomes of hemodynamically stable but high-risk patients were not defined.
Synthesis of the information on preincision IABP use 1-4 indicates that this practice favorably influences mortality and morbidity in CABG patients with myocardial ischemia or hemodynamic instability before the operation. A benefit for prophylactic preincision IABPs in high-risk patients requiring CABG who have no signs of hemodynamic instability was demonstrated by our study, but the benefit is less than in patients with preoperative hemodynamic instability. As more precise methods for predicting postoperative morbidity and mortality in CABG patients become available, the benefits of prophylactic preincision IABP use in appropriately selected patients are likely to increase.
A recent analysis of the Massachusetts Health Data Consortium database 16 showed wide variations among participating hospitals in risk-adjusted rates for IABP use at any time before, during, or after the CABG operation. This finding indicates that there is not currently a consensus among practicing surgeons regarding the indications for IABPs. Consensus regarding best practices in cardiac surgery has typically been achieved slowly. However, changes in the CABG operation during the past two decades, including an increase in the prevalence of internal thoracic grafting and uniformity in the dosing schedule for aspirin in CABG patients, show that surgeons reach consensus when there are data proving the benefit of a recommended practice.
Large multihospital studies, like the Alabama CABG Cooperative Project 5 and others, 17,18 provide a mechanism to evaluate the processes of care involved in cardiac surgery. Information from cooperative regional studies similar to the Alabama Project, accompanied by additional prospective and retrospective research to address specific questions, will allow surgeons to achieve consensus on optimal processes of care and move from consensus to implementation more rapidly than has previously been possible.
|
Disclaimer |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionDisclaimerReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Santarpino, F. Onorati, A. S. Rubino, K. Abdalla, S. Caroleo, E. Santangelo, and A. Renzulli Preoperative intraaortic balloon pumping improves outcomes for high-risk patients in routine coronary artery bypass graft surgery. Ann. Thorac. Surg., February 1, 2009; 87(2): 481 - 488. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Christenson Invited Commentary Ann. Thorac. Surg., September 1, 2008; 86(3): 827 - 827. [Full Text] [PDF]
|
|
|
|
 |
|
 C. Diez, R.-E. Silber, M. Wachner, M. Stiller, and H.-S. Hofmann EuroSCORE directed intraaortic balloon pump placement in high-risk patients undergoing cardiac surgery - retrospective analysis of 267 patients Interactive CardioVascular and Thoracic Surgery, June 1, 2008; 7(3): 389 - 395. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. T Christenson, J. Sierra, J.-A. Romand, M. Licker, and A. Kalangos Long Intraaortic Balloon Treatment Time Leads to More Vascular Complications Asian Cardiovasc Thorac Ann, October 1, 2007; 15(5): 408 - 412. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. R. Ramnarine, A. D. Grayson, W. C. Dihmis, N. K. Mediratta, B. M. Fabri, and J. A.C. Chalmers Timing of intra-aortic balloon pump support and 1-year survival Eur. J. Cardiothorac. Surg., May 1, 2005; 27(5): 887 - 892. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Aydin, O. Kandemir, S. Gunaydin, and Y. Zorlutuna A peeled off and entrapped intraaortic balloon catheter in the femoral artery: an unusual complication Interactive CardioVascular and Thoracic Surgery, June 1, 2004; 3(2): 314 - 316. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Dunning and B. Prendergast Which patients would benefit from an intra-aortic balloon pump prior to cardiac surgery? Interactive CardioVascular and Thoracic Surgery, December 1, 2003; 2(4): 416 - 419. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Dunning, J.K.K. Au, R.W.J. Millner, and A.J. Levine Derivation and validation of a clinical scoring system to predict the need for an intra-aortic balloon pump in patients undergoing adult cardiac surgery Interactive CardioVascular and Thoracic Surgery, December 1, 2003; 2(4): 639 - 643. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. F. Baskett, G. T. O'Connor, G. M. Hirsch, W. A. Ghali, K. Sabadosa, J. R. Morton, C. S. Ross, F. Hernandez, W. C. Nugent Jr, S. J. Lahey, et al. A multicenter comparison of intraaortic balloon pump utilization in isolated coronary artery bypass graft surgery Ann. Thorac. Surg., December 1, 2003; 76(6): 1988 - 1992. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Christenson Invited commentary Ann. Thorac. Surg., December 1, 2003; 76(6): 1992 - 1992. [Full Text] [PDF]
|
|
|
|
|
|
J. T Christenson Preoperative Intraaortic Balloon Pump for Salvage Myocardial Revascularization Asian Cardiovasc Thorac Ann, December 1, 2002; 10(4): 302 - 305. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Christenson, M. Cohen, J. J. Ferguson III, R. J. Freedman, M. F. Miller, E. M. Ohman, R. C. Reddy, G. W. Stone, and P. M. Urban Trends in intraaortic balloon counterpulsation complications and outcomes in cardiac surgery Ann. Thorac. Surg., October 1, 2002; 74(4): 1086 - 1090. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J.F. Baskett, W. A. Ghali, A. Maitland, and G. M. Hirsch The intraaortic balloon pump in cardiac surgery Ann. Thorac. Surg., October 1, 2002; 74(4): 1276 - 1287. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z. S. Meharwal and N. Trehan Vascular complications of intra-aortic balloon insertion in patients undergoing coronary reavscularization: analysis of 911 cases Eur. J. Cardiothorac. Surg., April 1, 2002; 21(4): 741 - 747. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Christenson and M. Cohen Preoperative IABP in high-risk patients reduces postoperative lactate release and subsequent mortality Ann. Thorac. Surg., March 1, 2002; 73(3): 1026 - 1027. [Full Text] [PDF]
|
|
|
|
 |
|
 W. L. Holman, R. M. Allman, M. Sansom, C. I. Kiefe, E. D. Peterson, K. J. Anstrom, S. S. Sankey, S. G. Hubbard, R. G. Sherrill, and for the Alabama CABG Study Group Alabama Coronary Artery Bypass Grafting Project: Results of a Statewide Quality Improvement Initiative JAMA, June 20, 2001; 285(23): 3003 - 3010. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
