doi: 10.1161/01.CIR.0000138198.62961.41
(Circulation. 2004;110:II-1 – II-6.)
© 2004 American Heart Association, Inc.
Surgery for Coronary Artery Disease |
On-Pump Versus Off-Pump Coronary Artery Bypass Surgery in a Matched Sample of Women
A Comparison of Outcomes
From HCA, Inc (M.J.M., P.B., F.H., M.K., A.K., S.B., L.T.), Nashville, Tenn; Cardiac Data Solutions, Inc (A.S.), Zionsville, Ind; and Rollins School of Public Health at Emory University (S.C., E.B.), Atlanta, Ga.
Correspondence to Michael J. Mack, MD, 7777 Forest Lane, Suite A323, Dallas, TX 75230. E-mail smullins{at}csant.com
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background— Women have consistently higher mortality and morbidity than men after coronary artery bypass grafting (CABG). Whether elimination of cardiopulmonary bypass and performance of coronary artery bypass grafting off-pump (OPCAB) have a beneficial effect specifically in women has not been defined.
Methods and Results— From January 1998 through March 2002, 21 902 consecutive female patients at 82 hospitals underwent isolated CABG, as reported in an administrative database. Propensity score computer matching was performed based on 13 variables representing patient characteristics and preoperative risk factors to correct for and minimize selection bias. A total of 7376 (3688 pairs) women undergoing CABG surgery were able to be successfully matched. In a propensity score computer-matched cohort, multivariate logistic regression (odds ratio) revealed that women undergoing on-pump surgery had a 73.3% higher mortality (P=0.002) and a 47.2% higher risk of bleeding complications (P=0.019).
Conclusions— In a retrospective analysis of women undergoing CABG, computer-matched to minimize selection bias, off-pump surgery led to decreased mortality and morbidity including bleeding complications.
Key Words: mortality • sex • cardiopulmonary bypass • women • surgery • administrative data
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Coronary artery disease is the leading cause of mortality in women, with 250 000 deaths annually, easily surpassing breast cancer as the leading cause.1 Furthermore, more women than men die each year of heart disease.2 Numerous studies have shown that women have a 1.5- to 2-times higher mortality rate after coronary artery bypass grafting (CABG) than men.3–12 Possible reasons for this increased mortality include a higher preoperative risk profile, including later clinical presentation,8 more acute presentation,5 older age,9 higher incidence of diabetes mellitus,7 higher incidence of left ventricular hypertrophy and hypertensive heart disease,1,13 and smaller coronary arteries.14,15
Off-pump CABG (OPCAB) has been introduced as a technique to lessen overall operative mortality and morbidity. Numerous series indicate that there may be some overall benefit to off-pump surgery,16–18 as well as a select benefit in certain high-risk subgroups.19,20 Two retrospective analyses have shown a decreased, but not statistically significant, mortality in women undergoing CABG off-pump compared with on-pump CABG.21,22 Because no randomized trial of OPCABG versus on-pump CABG in women exists, we retrospectively reviewed CABG outcomes in women in a large hospital system database. To minimize the effect of selection bias on outcomes, we used the technique of propensity score computer-matching of preoperative risk factors to obtain a valid comparison between the 2 treatment groups.23–27
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Study Population
All patients undergoing isolated CABG in the HCA, Inc hospital system are captured in the HCA Casemix Database. The HCA Casemix Database is a comprehensive administrative database containing patient, clinical, and outcome data on all cardiovascular patients at all HCA hospitals. Data are coded by International Classification in Diseases, 9th edition, Clinical Modification (ICD-9CM) codes from UB-92 billing information. There are 15 diagnostic codes (1 primary and 14 secondary codes) and 9 procedural codes (1 primary and 8 secondary) in the database. Data in this study were captured over a 17-quarter study period (51 months) from January 1, 1998 through March 31, 2002. All consecutive female patients undergoing isolated CABG in a total of 82 mainly community-based hospitals were included.
A propensity score computer-matched sample was used to create equivalent treatment groups. Initially, a total of 21 902 consecutive women patients at the 82 hospitals and discharged into 1 of the 3 primary CABG Diagnosis-Related Groups (DRGs) were identified. These DRGs are: DRG 106, CABG with percutaneous coronary intervention; DRG 107, CABG with cardiac catheterization; and DRG 109, CABG procedure only.
Of this cohort, 17 652 women (80.6%) underwent on-pump CABG and 4250 (19.4%) underwent OPCAB. Selection of procedure was performed was at the discretion of the individual surgeon. Determination of whether the procedure was performed on-pump or off-pump was by querying the ICD-9CM code for cardiopulmonary bypass from the UB-92 form. If the CABG procedure is performed on-pump, ICD-9CM code 39.61 (extracorporeal circulation auxiliary to pen heart surgery) is recorded. If the procedure is performed off-pump, this code is recorded as zero. If, however, a procedure was started off-pump and then converted, it would be recorded as "on-pump." There was a wide range in the number and percentage of off-pump procedures performed at the different sites. The average hospital performed 266 isolated CABG procedures on women during the study period.
Data Definitions
Patient characteristics, procedure complications, mortality, and hospital-processed characteristics were coded directly from the hospital discharge abstract (UB-92 form), which contains the principal diagnoses and procedures received by the patient. The definition in the ICD-9 were used to aggregate patient characteristics and procedure complications into categories.28 A total of 27 patient characteristics and risk factors and 13 patient outcomes were identified and compared using standard ICD-9 definitions.
Statistical Analysis and Propensity Score Calculation
The propensity score is the measure of the likelihood that a female patient would have undergone on-pump versus OPCABG using the patient’s covariate scores. Propensity scores were developed for each woman based on the 13 preoperative characteristics listed in Table 1. The C-statistic is 80.6%. To estimate the propensity scores, we fit a regression model in which the dependent variable was the log-odds of being an on-pump patient, with the covariates as described representing important patient and clinical characteristics. Using the results of the regression, we calculated a propensity score for all 21 902 females in the sample. Based on the propensity scores of the females receiving OPCAB, we matched their propensity score with an identical propensity score in the on-pump female CABG population. When >1 propensity score existed in the on-pump female CABG population, the match was randomly assigned to 1 of the matching female off-pump patients. All female CABG patients in the OPCAB group without a match on propensity score were excluded. Of the 4250 female patients undergoing OPCAB surgery, 3688 (86.8%) had a match to a female patient in the on-pump population. Thus, of a total of 7376 women undergoing CABG surgery, 3688 women undergoing OPCAB and 3688 women undergoing on-pump surgery were included in the final sample.
|
Table 2 shows the preoperative characteristics of the matched sample. Unadjusted means for women on-pump and off-pump are reported with P values calculated for each population using Student t tests.
|
Overall, of the 20 patient characteristics reported, only 5 conditions show statistically significant differences between the populations with on-pump versus those with OPCABG. They included older mean age and a higher incidence of insulin-dependent diabetes mellitus in the on-pump cohort. Women undergoing OPCAB had higher rates of chronic obstructive pulmonary disease and hypertension and previus percutaneous coronary interventions.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
By univariate analysis, the unadjusted mortality rate was lower, but not statistically significant, for women undergoing OPCAB surgery (3.17%) than for those undergoing on-pump CABG surgery (3.88%; P=0.101; Table 3) Women undergoing OPCAB surgery did have a significantly lower rate of major complications including shock/hemorrhage (2.68% versus 3.61%; P=0.023) and neurologic complications (1.03% versus 1.79%; P=0.006).
|
Of the other 8 minor complications, all these complications were lower or the same for women undergoing OPCAB surgery than for women undergoing on-pump CABG surgery. However, only 2 respiratory complications (2.47 versus 3.61, P=0.004) and adult respiratory distress syndrome (3.31 versus 4.64, P=0.003) were statistically significant.
On average, women undergoing OPCAB surgery were significantly more likely to have at least one arterial graft used than women undergoing on-pump CABG surgery (81% versus 72%; P=0.<001) and more total arterial grafting.
The length-of-stay from surgery to discharge for women undergoing OPCAB was no different, nor was the unadjusted cost ($25 068 OPCAB versus $25 140 CABG surgery, P=0.847). Women undergoing OPCAB surgery were discharged home more often than women who had undergone on-pump surgery (64.2% versus 59.1%; P
0.001), and the need for postdischarge healthcare was lower (Table 3).
Multivariate Regression
After controlling for patient, risk factors, and procedure characteristics, the estimated odds ratio (Table 4) indicates that women undergoing on-pump CABG surgery experienced a 73.3% higher mortality rate than women undergoing OPCAB (P=0.002).
|
Of the 13 complications analyzed, the estimated odds ratios for the variables, as reported in Table 2, indicate the use of CPB was statistically significant only for 1 complication, with women undergoing on-pump CABG surgery having a 35% higher likelihood of shock/hemorrhage complications (P=0.019). None of the other 12 complications was statistically significant.
Including total length of stay and including total cost in the regression analyses reveal that neither have any statistically significant differences for women in either the OPCAB group or the on-pump CABG surgery group.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
OPCAB grafting was reintroduced in the mid 1990s in an attempt to decrease the operative mortality and morbidity associated with conventional CABG. By elimination of cardiopulmonary bypass and the associated systemic inflammatory response associated, it was hoped that early adverse outcomes could be decreased without compromising long-term outcomes. Although 3 published randomized series of OPCAB grafting versus on-pump CABG have shown no clear benefit from the standpoint of operative mortality and neurologic complication, they are all underpowered, even in a meta-analysis, to demonstrate benefit.18,29,30 Significant benefit has been shown, however, in some other end points, such as postoperative atrial fibrillation and need for transfusion. Retrospective analysis of large databases have shown a statistically significant benefit in operative mortality with the use of OPCAB.16,17,31 However, all these analysis are subject to the selection bias that cannot be accounted for in any retrospective series.
Women consistently have a higher operative mortality than men undergoing CABG surgery. In the Society of Thoracic Surgeons National Cardiac Database, since 1994, women who comprised 28% of the entrants had a significantly higher operative mortality than men (4.5% versus 2.6%, P<0.001).31 Whether female gender is an independent predictor of adverse complications or whether there are other variables occurring more commonly in females that raise the risk has not been clear. The Bypass Angioplasty Revascularization Investigation (BARI) showed that after correction for the higher-risk profiles in women, female gender was actually an independent predictor of 5-year survival.4 Other possible reasons include women having a higher incidence of diabetes leading to increased complications, especially in the insulin-dependent diabetic group.32 Younger women have been demonstrated to have higher hospital mortality rates than men, especially when younger than 50 years of age.6 Later presentation with urgent or emergent status are also variables occurring more common in women, leading to higher complication rates.5,8 More recently, a higher incidence of left ventricular hypertrophy and hypertensive heart disease in women has been proposed as causes of higher complications.1
There have been 2 previous studies that have specifically examined the possible benefit of off-pump in women. An analysis of the female subgroup by Petro et al21 showed that there was lower, but insignificant, mortality in the off-pump group compared with on-pump surgery (2.3% versus 4.1%, P=0.12). In an earlier analysis by our group, Brown et al22 also showed an insignificantly lower mortality in females undergoing off-pump compared with on-pump (3.12% versus 3.9%, P=0.052). Randomized studies are generally considered the highest level of evidence for comparing the treatment with a control.27 The essential feature of the randomized study is that some random mechanism controls treatment assignment, giving each patient the same probability of receiving the treatment and protecting against a biased comparison caused by patient selection.
On retrospective analyses, because selection bias may play a significant role in selection for patients undergoing OPCABG, meaningful analysis is difficult. We have attempted to control for selection bias by using propensity score comparison to create cohorts with equal preoperative risk factors. Propensity score analysis was constructed to minimize the role of selection bias in choosing the treatment received by the individual woman. In comparing females receiving on-pump versus OPCABG, 1 group is chosen as the treatment group, eg, on-pump, and the other is the control. Then the propensity score is simply the probability that a patient given her particular set of characteristics received on-pump CABG. A group of patients with the same propensity score are thus equally likely to have been assigned to the on-pump treatment group and equally likely to have been assigned to OPCABG. The effect is the same as if they had been randomly allocated to either treatment regardless of which treatment they actually received. According to Grunkemeier, it is "randomization after the fact."27
Using multivariate logistic regression analysis of the propensity score-matched sample, we found that women undergoing OPCABG surgery had a significantly lower mortality rate and experienced fewer bleeding complications. We also found that there was no benefit in other major and minor complications or in length of stay or cost. By regression analysis controlling for other covariates, the cost difference between women undergoing off-pump and on-pump CABG was $85, which was not statistically significant.
Although demonstrating improved mortality, this study does not clarify why OPCAB may benefit women. OPCAB has been shown in other series to specifically benefit patients with significantly higher risk and other comorbidities. Although women clearly fall into a higher risk category with operative mortality consistently higher than men, the mechanism of improved outcomes is not answered by this study because the risk profile is essentially the same between the matched groups. However, our data support a specific and significant benefit to off-pump surgery in women.
Limitations
There are some important limitations to this study that mandate cautious interpretation. First, the HCA Casemix Database is an administrative database and lacks particular clinical insights that might be useful in segmenting women into severity levels of coronary heart disease. Second, the data are not risk-adjusted for specific clinical characteristics, and it is possible that a physician’s choice of procedure is based on clinical differences that are not captured by an administrative database. Third, intention to treat cannot be identified. The database only captures whether the pump was used and not what the intention at the onset of the procedure was. Thus, conversions to off-pump versus on-pump are captured in the on-pump group. This conceivably could bias outcomes adversely in the on-pump group. However, in previous analyses, we have determined that the typical conversion rate (6% to 8%) and mortality in converted patients (4% to 6%) are not sufficiently significant to affect the overall outcomes. Lastly, the female CABG patient population represents pairs of women matched based on propensity scores. Although the matching pattern is random, the underlying female population does not reflect a random population.
|
Footnotes |
|---|
Presented at American Heart Association Scientific Sessions, November 9–12, 2003, Orlando, Florida.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
- Wenger NK. Is what’s good for the gander good for the goose? J Thorac Cardiovasc Surg. 2003; 126: 929–931.
[Free Full Text]
- Am Heart Association Web site. Available at: www.women.americanheart.org.
- Edwards FH, Carey JS, Grover FL, et al. Impact of gender on coronary bypass operative mortality. Ann Thorac Surg. 1998; 66: 125–131.
[Abstract/Free Full Text]
- Jacobs AK, Kelsey SF, Brooks MM, et al. Better outcome for women compared with men undergoing coronary revascularization. A report from the Bypass Angioplasty Revascularization Investigation (BARI). Circulation. 1998; 98: 1279–1285.
[Abstract/Free Full Text]
- O’Rourke DJ, Malenka DJ, Olmstead EM, et al. for the Northern New England Cardiovascular Disease Study Group. Improved in-hospital mortality in women undergoing coronary artery bypass grafting. Ann Thorac Surg. 2001; 71: 507–511.
[Abstract/Free Full Text]
- Vaccarino V, Abramson JL, Veledar E, et al. Sex differences in hospital mortality after coronary artery bypass surgery. Evidence for a higher mortality in younger women. Circulation. 2002; 105: 1176–1181.
[Abstract/Free Full Text]
- Vaccarino V, Lin ZQ, Kasl SV, et al. Gender differences in recovery after coronary artery bypass surgery. J Am Coll Cardiol. 2003; 41: 307–314.
[Abstract/Free Full Text]
- O’Connor GT, Morton JR, Olmstead EM, et al. For the Northern New England CV Study Group. Excess mortality among women undergoing coronary artery bypass graft surgery [abstract]. Circulation. 2002; 106 (suppl1): II-552.
- Koch CG, Mangano CM, Schwann N, et al. Is it gender, methodology, or something else? J Thorac and Cardiovasc Surg. 2003; 126 (4): 932–935.
[Free Full Text]
- Christakis G, Weisel R, Buth K, et al. Is body size the cause for poor outcomes of coronary artery bypass operations in women? J Thorac Cardiovasc Surg. 1995; 110: 1344–1358.
[Abstract/Free Full Text]
- Woods S, Noble G, Smith J. et al. The influence of gender in patients undergoing coronary artery bypass graft surgery: an eight-year prospective hospitalized cohort study. J Am Coll Surg. 2003; 196: 428–434.[CrossRef][Medline]
[Order article via Infotrieve]
- Lawton JS, Brister SJ, Petro KR, et al. Surgical revascularization in women: Unique intraoperative factors and considerations. J Thorac and Cardiovasc Surg. 2003; 126: 936–938.
[Free Full Text]
- Mickelborough LL, Takagi Y, Mariyama H, et al. Is sex a factor in determining operative risk for aortocoronary bypass surgery? Circulation. 1995; 92 (suppl 1): II80–II84.[Medline]
[Order article via Infotrieve]
- Tyras DH, Barner HB, Kaiser GC, et al. Myocardial revascularization in women. Ann Thorac Surg. 1978; 25: 449–453.[Abstract]
- Mickleborough LL, Carson S, Ivanov J. Gender differences in quality of distal vessels: effect on results of coronary artery bypass grafting. J Thorac Cardiovasc Surg. 2003; 126: 950–958.
[Abstract/Free Full Text]
- Mack MJ. Current Status and Outcomes of Coronary Revascularization 1999–2002: 154,701 Consecutive Surgical and Percutaneous Coronary Interventions. Ann Thorac Surg. 2004; 77: 761–768.
[Abstract/Free Full Text]
- Mack MJ, Pfister A, Bachand D, et al. Comparison of coronary bypass surgery with and without cardiopulmonary bypass in patients with multivessel disease. J Thorac Cardiovasc Surg. 2004; 127: 167–173.
[Abstract/Free Full Text]
- Puskas JD, Williams WH, Duke PG, et al. Off-pump coronary artery bypass grafting provides complete revascularization with reduced myocardial injury, transfusion requirements, and length of stay: A prospective randomized comparison of two hundred unselected patients undergoing off-pump versus conventional coronary artery bypass grafting. J Thorac Cardiovasc Surg. 2003; 125: 797–808.
[Abstract/Free Full Text]
- Stamou SC, Corso PJ. Coronary revascularization without cardiopulmonary bypass in high-risk patients: a route to the future. Ann Thorac Surg. 2001; 71: 1056–1061.
[Abstract/Free Full Text]
- Ascione R, Narayan P, Rogers CA, et al. Early and midterm clinical outcome in patients with severe left ventricular dysfunction undergoing coronary artery surgery. Ann Thorac Surg. 2003; 76: 793–799.
[Abstract/Free Full Text]
- Petro KR, Dullum MK, Garcia JM, et al. Minimally invasive coronary revascularization in women: a safe approach for a high-risk group. Heart Surg Forum. 2000; 3: 41–46.[Medline]
[Order article via Infotrieve]
- Brown PP, Mack MJ, Simon AW, et al. Outcomes experience with off-pump coronary artery bypass surgery in women. Ann Thorac Surg. 2002; 74: 2113–2119.
[Abstract/Free Full Text]
- D’Agostino RB. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998; 17: 2265–2281.[CrossRef][Medline]
[Order article via Infotrieve]
- Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983; 70: 41–55.
[Abstract/Free Full Text]
- Drake C, Fisher L. prognostic models and the propensity score. Int J Epidemiol. 1995; 24: 183–187.
[Abstract/Free Full Text]
- Rubin D. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997; 127: 757–763.
[Abstract/Free Full Text]
- Grunkemeier GL, Payne N, Jin R, et al. Propensity score analysis of stroke after off-pump coronary artery bypass grafting. Ann Thorac Surg. 2002; 74: 301–305.
[Free Full Text]
- Practice Management Information Corporation. ICD-9-CM: International Classification of Diseases: 9th Revision, Clinical Modifications: 6th Edition. Los Angeles, CA. Practice Management Information Corporation, 2004.
- Van Dijk D, Nierich AP, Jansen EW, et al. Early outcome after off-pump versus on-pump coronary bypass surgery: results from a randomized study. Circulation. 2001; 104: 1761–1766.
[Abstract/Free Full Text]
- Angelini GD, Taylor FC, Reeves BC, et al. Early and midterm outcome after off-pump and on-pump surgery in Beating Heart Against Cardioplegic Arrest Studies (BHACAS 1 and 2): a pooled analysis of two randomized controlled trials. Lancet. 2002; 359: 1194–1199.[CrossRef][Medline]
[Order article via Infotrieve]
- Society of Thoracic Surgeons. STS National Database Executive Summary. Available at http://ctsnet.org/file/stsnationaldatabasefall2002executive summary.pdf 2003.
- Szabo Z, Hakanson E, Svedjeholm R. Early postoperative outcome and medium-term survival in 540 diabetic and 2239 nondiabetic patients undergoing coronary artery bypass grafting. Ann Thorac Surg. 2002; 74: 712–719.
[Abstract/Free Full Text]
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||