Author + information
- Received January 20, 2006
- Revision received January 31, 2006
- Accepted February 7, 2006
- Published online May 16, 2006.
- ↵⁎Reprint requests and correspondence:
Dr. Robert H. Jones, Duke Clinical Research Institute, P. O. Box 17969, Durham, North Carolina 27715.
A comprehensive review of articles on cardiovascular surgery in major general and cardiovascular journals published in 2005 identified a large number that were scientifically sound and broadly relevant. The scientific quality of the 43 articles chosen as most likely to affect the future practice of cardiovascular specialists made it difficult to identify a single work as the “outstanding cardiovascular surgical article of 2005.” However, the “most valuable topic award” clearly goes to four articles that distill down and validate multivariable model information into bedside risk prediction tools that quantitate:
• The natural history of patients with prior coronary artery bypass grafting (CABG) using coronary angiography.
• Operative mortality and five-year survival in patients with low ejection fraction (EF) using five clinical variables.
• Major infection after CABG using preoperative variables with or without intraoperative variables.
• Operative risk for aortic and mitral valve surgery using clinical information.
Careful use of these scores to adjust estimates for differences in outcomes in care environments that develop (and care providers who apply) the predictive tool will infuse quantitative thinking into physician deliberation and provide patients with information describing personalized risks and benefits to inform their deliberations about an operation.
Coronary artery disease
Delivery of High-Quality CABG Care
Influence of CABG volume on patient outcomes
Clinical data on 27,355 adults voluntarily entered into the California CABG Mortality Report Program between 1997 and 1998 were used to examine the influence of hospital-specific case volume rates on patient mortality (1). Annual isolated CABG operations ranged from a low of 44 to a high of 1,283 among the 68 participating hospitals. The multivariable logistic regression model that was used to adjust for baseline patient differences using 22 clinical variables performed with a C index of 0.80 in a validation cohort. Adjusted in-hospital patient mortality rates among the three cohorts stratified into low, medium, and high tertiles of hospital case volumes were displayed as a function of percentile of projected patient risk (Fig. 1).The 44 low-volume hospitals operated on a higher risk population, with an expected mortality of 3.25% compared with 2.37% in the 19 medium-volume hospitals and 1.97% in the high-volume hospitals (Table 1).Adjusting for individual patient-expected risk alone reduced facility level of variance by 38%; adding hospital volume categories to the regression model reduced variance by an additional 61%. Therefore, 99% of the total variance in outcomes across facilities was explained by these two factors. As a group, low-volume hospitals operated less well on more high-risk CABG patients than hospitals with larger volumes, and these two related factors resulted in 157 excess deaths (0.57%) in the total 27,355-patient cohort.
Mandatory statewide CABG reporting in California supported by the California Society of Thoracic Surgeons (STS), the California Chapter of the American College of Cardiology, and the California Medical Association provided data to compare with the statewide New York CABG experience during 1999 to 2001 (2). Coronary artery bypass mortality was 33% higher in California than New York. After adjustment for differences in baseline patient characteristics, 705 excess deaths (29%) of the 2,431 total CABG deaths in California are explained solely by the greater prevalence of low-volume hospitals in California (3).
If the U.S. rate of excess CABG mortality attributable to low-volume CABG delivery has now decreased to one-half the California rate reported in this study (0.5 × 0.0029), and only 200,000 CABG operations are performed each year, then 290 potentially preventable CABG deaths occur throughout the U.S. each year (0.0029 × 0.5 × 200,000), which is equivalent to an annual fatal crash of a large airliner.
Administrative data were used to compare hospital or 30-day mortality of percutaneous coronary intervention (PCI) and CABG in 42,737 Medicare beneficiaries served by 15 cardiac specialty hospitals and 82 general hospitals during 2000 and 2001 (3). The mean annual CABG volume in Medicare patients was 118 in general hospitals versus 285 in specialty hospitals (p < 0.001). The multivariate model used to adjust for fewer coexisting conditions and fewer acute myocardial infarctions (MIs) among patients treated in specialty hospitals had a C statistic of 0.74. The adjusted odds ratio (OR) for death after CABG remained lower in specialty than general hospitals (0.084, p = 0.05). That excess mortality in general hospitals related to lower case volumes was confirmed by a case volume stratified analysis adjusted for baseline patient characteristics that showed no difference in CABG mortality among specialty and general hospitals with similar case volumes.
A high CABG case volume maintains team skill among care providers (just as it does for athletic teams) and explains the mechanism responsible for the relationship between high CABG volume and low mortality. A recent decrease in total patients undergoing CABG, resulting in an excess of cardiac surgeons ready to provide surgical coverage for PCI in small hospitals, now fuels a trend toward more small-volume CABG programs. Broader patient access to primary PCI, often used to justify this trend, should instead be achieved by improving emergency cardiovascular care systems in the U.S. Our cardiovascular profession must assume responsibility for reducing preventable loss of life by making our care delivery systems optimally safe.
Cost of CABG
CABG costs were $20,672 ± $241 in patients operated at five U.S. hospitals compared with $10,373 ± $123 for 7,319 CABG patients at four Canadian hospitals (4). U.S. patients were older, were more likely to be women, and were discharged from the hospital sooner than Canadian patients. After correction for these baseline differences, mortality did not differ significantly between the two countries, but CABG performed in the U.S. had an 82.5% higher cost than if performed in Canada (p < 0.001). The higher U.S. cost was driven by higher direct and overhead costs in U.S. hospitals for all expenses except nursing care (Fig. 2).In this population, women undergoing CABG had a $2,809 greater cost in the U.S. and $1,057 greater cost in Canada than men (p < 0.0001) (5). After adjusting for age and comorbid conditions, female gender remained associated with a 10% increase in length of stay (p = 0.0001), a 97% increase in mortality (p = 0.0006), and a 7% increase in overall cost (p < 0.0001).
The influence of bypass graft patency and extensiveness of intrinsic coronary atherosclerosis on subsequent survival without repeat revascularization was assessed in 3,178 patients with previous CABG using the date of diagnostic catheterization as time zero and censoring patients at any subsequent cardiac revascularization over a mean interval of 3 years after catheterization (6). A graft index expressed as a 0 to 100 number relating directly to observed prognostic importance proved to be the third most important predictor of subsequent death, and, along with age and presence of heart failure, made up 50% of the total prognostic information (Table 2).This graft index was first derived from descriptions of disease in native vessels and whether myocardial territories jeopardized by intrinsic coronary disease were served by a patent bypass graft (<75% stenosis). When combined with information describing presence or absence of a ≥75% left main stenosis and whether an intact graft conduit was an internal mammary artery and not a vein, this extended coronary bypass graft index identified patients with a five-year survival that ranged from a low of 50% (patients with all previous bypass grafts significantly diseased and three-vessel and left main stenosis) to a high of 95.7% (patients with patent native coronary arteries) (Table 3).This coronary bypass graft index provides the first quantitative description of the prognostic content of coronary anatomy information in patients with prior CABG.
The influence of race on operative and 12-year outcomes was retrospectively evaluated in a consecutive cohort of 304 African-American and 6,073 Caucasian patients undergoing isolated CABG at a single institution between 1991 and 2003 (7). Logistic regression and Cox multivariable models were used to adjust for baseline differences in patient characteristics, and propensity scores were used to match the African-American cohort with a similar Caucasian cohort. Unadjusted and adjusted operative mortality was similar in African-American and Caucasian patients. Propensity-matched cohorts of African-American and Caucasian patients also had similar long-term survival (Fig. 3).However, Medicaid status, which is prevalent among African Americans, was significant but contributed only about 1% of the prognostic information in a Cox model that was dominated by patient age and presence of renal failure and/or diabetes.
A retrospective study of a cohort of 566,785 white and 24,354 black Medicare beneficiaries undergoing CABG in 1,091 hospitals from 1997 to 2000 confirmed that, after adjustment only for patient characteristics, mortality at 30, 90, and 365 days was 8%, 11%, and 25% higher, respectively, in black than in white patients (8). After adjustment for hospital effects, the 30- and 90-day mortalities were similar, but mortality at 365 days was 70% higher in black patients. However, zip code data was the only information available to reflect socioeconomic differences. Socioeconomic status appears to be a more dominant determinant of CABG outcome than race.
Adding CABG to intensive medical therapy may increase long-term survival benefit to patients with ischemic cardiomyopathy but at a higher procedure mortality than in patients with normal left ventricular (LV) function. Demonstration of myocardial viability in an underperfused and dysfunctional myocardial region is widely used to identify patients for revascularization despite lack of evidence for its discriminatory power (9). Until this question is definitively addressed by the ongoing STICH trial, balancing CABG operative risk and survival benefit against quantitative estimates of survival in similar medically treated patients provides the best objective information to counsel patients with ischemic cardiomyopathy about the value of revascularization.
Stratification of 55,515 patients from the New York State CABG registry between 1997 and 1999 into four EF groups demonstrated the influence of LV dysfunction on in-hospital and early CABG mortality (10) (Fig. 4).Multivariable analysis in the 2,442-patient subgroup (4.4%) of this cohort with EF ≤20% and an in-hospital mortality of 6.5% identified older age, MI within 6 h, reoperation, renal failure, and hepatic failure as strong additional risk predictors.
A retrospective single-institution study evaluated the long-term predictors of survival during a mean follow-up of 4.1 years in 544 consecutive patients with an EF ≤25% who underwent CABG from 1992 to 2002 (11). A multivariable Cox regression analysis permitted construction of a predictive score for long-term mortality in 360 patients that was validated in a set of 184 patients. The five variables included in the multivariable equation created the acronym HAVOC for the score, based upon H – heart failure, A – age, V – vascular disease, O – operation as emergency, C – chronic obstructive pulmonary disease using the New York Cardiovascular Database definitions (Table 4).The four-level HAVOC score provides estimates of 30-day CABG mortality and subsequent five-year survival. A low HAVOC score of 1 predicts a 3% 30-day CABG mortality and an 82.3% five-year survival, whereas a high HAVOC score of 4 predicts a 10.9% 30-day CABG mortality and a 45.5% five-year survival. This quantitative tool for assessment of surgical outcomes in patients with low EF when compared with other estimates of risks and benefits expected from medical treatment alone should assist individual patients and their physicians in deciding whether or not to undergo revascularization for this high-risk condition.
Major infection after CABG increases perioperative morbidity and long-term mortality, and no robust model exists to predict its occurrence. A search of the STS database identified major infection (septicemia 35%, vein harvest site 33%, mediastinitis 25%, other 7%) to have occurred in 11,636 (3.5%) of 331,429 CABG operations performed from January 2002 to December 2003 (12). Logistic regression was used to identify risk factors and to generate and validate two predictive models. One model included only information known preoperatively. A second model included both preoperative and intraoperative characteristics. These two risk scores were developed by simplifying multivariable analysis equations when applied to the validation sample performed with a C index of 0.69 when based only on preoperative characteristics and 0.70 when based on pre- and intraoperative characteristics (Table 5).The capacity of this tool to predict major CABG infection rates that ranged from 1% to 16% is certain to be valuable for more accurately informing patients of their infection risk and to focus intensive use of preventive strategies to patients predicted to have high perioperative infection risk.
The common use of clopidogrel in patients with acute coronary syndrome assures that decisions must often be made about its use in patients for whom CABG may later be considered as a therapeutic option. The effect of preoperative clopidogrel use on perioperative bleeding was evaluated in 2,359 patients undergoing isolated CABG at a single center between January 2000 and June 2002 (13). In this observational study, 415 patients (17.6%) received clopidogrel before CABG and 1,944 (82.4%) did not. Clopidogrel use was a strong independent predictor in multivariable modeling of all end points related to bleeding but did not significantly increase operative mortality. Propensity-score matched analysis showed that clopidogrel administration increased reoperation for bleeding more than four-fold and transfusion of any blood, multiple units of blood, or platelets by two-fold. These data strongly advocate for a management strategy of potential CABG candidates with acute coronary syndrome that delays clopidogrel administration until coronary angiogram information is available to guide the choice of PCI or CABG if revascularization is indicated.
The influence of careful hyperglycemic control during cardiopulmonary bypass was evaluated in 1,579 diabetic and 4,701 non-diabetic patients undergoing cardiac surgery at a single institution between 1999 and 2001 (14). Stepwise multivariate logistic regression analysis was used to correct for differences in baseline characteristics to assess the importance of keeping peak glucose levels at less than 20 μmol/l (360 mg/dl) during cardiopulmonary bypass. A high glucose level during cardiopulmonary bypass was an independent predictor of in-hospital mortality in non-diabetic patients (OR 1.12, p < 0.0001) and in diabetic patients (OR 1.20, p 0.0005). These data argue for careful glucose control during cardiopulmonary bypass.
Postoperative assessment and management
A MEDLINE database search of English-language publications during the years 1966 to 2004 yielded only 14 randomized clinical trials evaluating drug use in 100 or more patients undergoing CABG that examined outcomes more than one year after CABG (15).
Aspirin was evaluated in eight randomized clinical trials that included 147 to 777 patients treated with at least 320 mg of aspirin per day with or without dipyridamole. Aspirin appears to be effective in reducing bypass graft occlusion at 12 months after CABG when administered within one day after the operation. Evidence is lacking to support the benefit of aspirin in reducing the incidence of cardiovascular events more than one year after CABG.
Three randomized, controlled trials of antilipid agents in 1,934 patients suggested that long-term treatment with antilipid agents prevented progression of both native coronary artery and coronary graft atherosclerosis and subsequent cardiovascular events in CABG patients. None of these studies started treatment at the time of surgery, and only one of the three trials examined a statin drug.
A single trial of beta-blockers randomized 967 patients to receive or not receive a metoprolol regimen begun within three weeks after CABG and continued for two years. No adverse end point was found to be significantly different between the two groups at two-year follow-up.
A single calcium-channel blocker trial randomized 120 CABG patients with normal perfusion of the radial artery after one year of treatment with 120 mg daily or oral diltiazem to either continue or suspend further treatment. Four-year follow-up showed no significant difference in ischemic end points.
Only one randomized clinical trial examined the benefit of angiotensin-converting enzyme inhibition after CABG. The QUO VADIS study evaluated administration of quinapril 40 mg daily, which had no effect on the primary end point of change in total exercise duration or the incidence of ischemia on Holter monitoring. However, 3.5% of patients receiving quinapril experienced ischemic events such as angina, death, MI, repeat revascularization, stroke, or transient ischemic attack versus 15% of patients receiving placebo (p = 0.02).
This paper exposes the weakness of the evidence base available to guide postoperative therapy in the CABG patient. The growing tendency to incorporate medication use in CABG patients at time of post-CABG discharge as a quality indicator for an entire institution is not only unsubstantiated by rigorous evidence but potentially dangerous to patients. Drugs clearly indicated in patients with coronary artery disease, such as angiotensin-converting enzyme inhibitors, beta-blockers, and lipid-lowering agents, provide only modest survival benefit per month but may impose substantial risk when started in the early postoperative period, when CABG patients are still recovering from the psychologic and physiological stress of surgery associated with high circulating catecholamines, hemodilution anemia, pain medication side effects, sleep and diet disruption, and general elevation of daily stress responses.
An adequately powered Phase III, double-blind, placebo-controlled trial (PREVENT IV) studied the efficacy and safety of edifoligide, a E2F transcription factor decoy, for prevention of vein graft failure in 3,014 patients undergoing primary CABG (16). Vein grafts were treated ex vivo with either edifoligide or placebo in a pressure-mediated delivery system. A total of 1,920 patients (80%) reached the end point of either death (n = 91) or vein graft patency assessment by 12- to 18-month follow-up angiography (n = 1,829). Edifoligide had no primary effect on per patient graft conduit occlusion, which was 45.2% in the edifoligide group versus 46.3% in the placebo group, on any secondary angiographic end point, or on the incidence of major cardiac events at one year (6.7% in the edifoligide group vs. 8.1% in the placebo group). At one year among the 1,593 patients who had thoracic artery angiography, internal thoracic artery grafts had a lower failure rate than vein grafts (8% vs. 29%). Despite its negative primary outcome, this trial remains highly significant as the first adequately powered prospective evaluation of a clinically important hypothesis related to the intraoperative conduct of CABG.
On or off cardiopulmonary bypass
A meta-analysis of 37 randomized clinical trials comparing off- and on-pump CABG in 3,369 patients showed no significant differences in short-term mortality or major morbidity (17). End point differences suggested that 85,000 patients would need to be randomized to address the end point of 30-day mortality with an 80% power. However, off-pump CABG significantly decreased atrial fibrillation (AF), transfusion, and other postoperative complications. Only 20 of these 37 studies reported conversion rates, which averaged 8% of off-pump CABG patients converting to on-pump and 1.7% of on-pump CABG patients converting to off-pump. However, no study reported outcomes for the patient subset requiring conversion. Subsequent analysis by a single-center randomized comparison of off- or on-pump CABG reported that 20 of the 300 randomized patients (6.7%) converted from off- to on-pump CABG (18). This conversion subgroup had a 10% operative mortality as well as high rates of morbid events.
Another meta-analysis of off-pump CABG surgery also reported results of 37 randomized clinical trials but included six different studies from the prior meta-analysis and also included separate meta-analysis of 22 observational studies in 293,617 patients (19). In randomized clinical trials, off-pump CABG was associated with reduced AF (OR 0.59; 95% confidence interval [CI] 0.46 to 0.77) and non-significant trends toward reduced 30-day mortality, stroke, and MI. Observational studies showed off-pump CABG to be associated with reduced 30-day mortality (OR 0.72; 95% CI 0.66 to 0.78), stroke (OR 0.62; 95% CI 0.55 to 0.69), MI (OR 0.66; 95% CI 0.50 to 0.88), and AF (OR 0.77; 95% CI 0.74 to 0.82). At one to two years, off-pump CABG was associated with trends toward reduced mortality as well as increased repeat revascularization.
Therefore, randomized controlled trials confirmed the reduction of AF reported in observational studies but did not find reductions in short-term mortality and morbidity demonstrated by observational studies. These major differences resulted primarily from randomized clinical trials being reported by intention to treat, whereas observational studies typically report outcomes of the operation received and thereby overstate the effectiveness of off-pump CABG.
The Merged Cardiac Registry was used to evaluate the risk of conversion from off- to on-pump CABG in 70,514 isolated CABG procedures performed from January 1998 to March 2004 in 40 facilities. Cardiopulmonary bypass was begun and completed on pump in 62,634 patients and begun off pump in 7,880 patients, of whom 7,424 patients (94.2%) were completed as intended and 456 (5.8%) were converted to on-pump (20). A logistic regression model of factors related to conversion from off- to on-pump CABG included need for urgent procedure, LV hypertrophy, Canadian Cardiovascular Society Class III or IV angina, prior CABG, and history of cerebrovascular disease. A higher mortality and morbidity were observed in patients requiring conversion compared with the on-pump cohort and the residual cohort defined by intention to treat by off-pump CABG (Table 6).Analyzing patients by intention to treat showed no significant difference in risk-adjusted operative mortality, but analyzing the failed off-pump patients separately or as part of the cardiopulmonary bypass cohort made the off-pump CABG appear to have a significantly lower operative mortality. These data illustrate the importance of intention-to-treat analysis when comparing these two modes of revascularization by CABG.
A well-designed prospective observational study was conduced by the Canadian Off-pump CABG Registry Investigators among 13 high-volume cardiac centers and 42 surgeons across Canada between March 12, 2002, and December 31, 2002 (21). The time and reason for conversion from off-pump to on-pump were prospectively recorded when any patient received cardiopulmonary bypass during an intended off-pump CABG. Standard definitions were prospectively applied uniformly by all centers. Telephone follow-up at one year was complete in 97.7% of all patients. A prospective sample size of 1,600 patients was identified to be sufficient to detect a 1.5% decrease in one-year mortality at a reasonable level of significance. A stepwise multivariable logistic regression analysis was performed to identify a population of patients with a 1:1 ratio of benefit anticipated from off- or on-pump CABG. Over 20 months ending in December 2002, 3,250 consecutive isolated CABG patients (1,656 off-pump and 1,693 on-pump) were entered into the study and reported by intention to treat. More bypass grafts were completed in on-pump patients. No significant differences were observed in the rates of in-hospital mortality or stroke, but MI and acute renal disease were higher in off-pump patients (Table 7).At one year, the only significant difference was a lower stroke rate in the off-pump cohort. No preoperative characteristics clearly defined subgroups most likely to benefit from a specific operative approach.
The exemplary methodology of this study serves as a template for the design of any future prospective observational study. However, the 0.2% difference in-hospital favoring off-pump CABG and the 0.4% difference in one-year mortality favoring on-pump CABG suggest that a very large sample size will be needed to show differences with statistical certainty. Such a trial would be feasible only with a large national registry such as the STS database.
Need for concomitant mitral repair for insufficiency
A retrospective review of 251 patients with 3+ ischemic mitral regurgitation undergoing CABG between 1991 and 2001 included 31 patients who had concomitant mitral annuloplasty performed at the discretion of the operating surgeon (22). Parametric multiphase hazard regression models were used to correct for differences in other baseline and operative characteristics. The performance of a mitral annuloplasty at the time of CABG reduced the degree of mitral regurgitation but did not predict greater long-term survival.
Postoperative Assessment and Management
Uniform prospective electrocardiography and creatine kinase-MB fraction (CK-MB) was monitored in 800 CABG patients who had participated in the PREVENT IV study to reflect myocardial injury at CABG for predicting later death or LV dysfunction (23). Multiple logistic regression models confirmed increased cardiopulmonary bypass time to be the strongest predictor of 30-day mortality or early severe LV dysfunction. Peak CK-MB ≥100 ng/ml and age were the other two significant variables. Q-wave MI was not an independent predictor of subsequent death or severe LV failure. These are the first data to show prospective monitoring of CK-MB after CABG to add independent prognostic information after clinical and electrocardiographic outcome markers are considered.
Operative stroke influence on survival
The Northern New England Cardiovascular Disease Study Group examined the influence of stroke (defined as a new fixed neurologic defect persisting at least 24 h after surgery) on 10-year all-cause mortality survival derived from the National Death Index on 35,733 consecutive patients undergoing isolated CABG surgery from 1992 to 2001 (24). Stroke occurred in 575 (1.61%) of the 35,733 patients undergoing isolated CABG. Patients with strokes were about five years older, more likely to be women; slightly smaller; and have lower EFs, more comorbidities, three-vessel disease, diabetes, renal disease, and urgent need for surgery than patients who did not sustain a stroke. The annual incidence of death over 10 years was 3.86 per 100 person-years in the entire cohort. After adjusting for the differences in baseline patient characteristics, the hazard ratio for increased death in patients sustaining a stroke at CABG was 3.2 (p < 0.001). The unadjusted survival curve defines the actual subsequent death burden incurred by patients who sustain an early postoperative stroke. These data offer a quantitative combined end point of death and predicted death at a subsequent time that should prove useful in future CABG outcome studies, because not only is the added death burden real, but the stroke also reduces quality of life remaining before this hastened death (Fig. 5).
Duration of AF therapy
From a consecutive cohort of 1,206 patients, the 129 patients in whom new AF after CABG successfully reverted to sinus rhythm in the postoperative period were prospectively randomized at discharge to receive anti-arrhythmic therapy for one, three, or six weeks (25). Beta-blockers were continued in all patients receiving these agents before surgery. Discontinuation of the specific medication used to treat the perioperative episode of AF (most commonly amiodarone and beta-blockers) did not lead to significant difference in recurrence of AF among the three groups. These data suggest that anti-arrhythmic therapy for perioperative AF for as short as one week may be appropriate.
Comparison of CABG and PCI
Meta-analysis of recent trials
The primary clinical end point of death, MI, and stroke at one year was reported from a database composed of individual data on 1,533 patients enrolled in four randomized trials comparing CABG and PCI with bare metal stenting (Arterial Revascularization Strategies Study (ARTS), Stent or Surgery [SOS] trial, Argentine Randomized Trial of PCI Versus CABG in Multivessel Disease 2 [ERACI-2], and Medicine, Angioplasty, or Surgery Study 2 [MASS-2]) (26). The primary end point of death, MI, or stroke at one year did not differ between a stenting or CABG revascularization strategy, but repeat revascularization procedures occurred more frequently in patients allocated to stenting (18% vs. 4.4%, p < 0.001) (Fig. 6).
The ARTS trial 5-year outcomes
During five-year follow-up of the 1,205 patients in the ARTS trial, death occurred in 48 (8.0%) of 600 patients with multi-vessel disease randomized to receive bare-metal stents and in 46 (7.6%) of the 605 patients randomized to CABG (p = 0.83) (27). Among 208 diabetic patients, mortality was 13.4% in the stent group and 8.3% in the CABG group (p = 0.27). The incidence of repeat revascularization was 30.3% in the stent group and 8.8% in the CABG group (p < 0.001).
New York outcomes
Three-year death and subsequent revascularization events were compared in 59,314 patients with multi-vessel disease who underwent either CABG (n = 37,212, 65%) or PCI with stenting (n = 22,102, 35%) in New York state between January 1, 1997, and December 31, 2000 (28). Rates of adverse outcomes were adjusted by proportional hazard methods to account for differences in baseline markers of severity of illness. The median follow-up was 1.9 years in the CABG group and 1.6 years in the PCI group. After adjustment for severity of illness, CABG was associated with a significantly higher likelihood of survival in all anatomic subgroups (Fig. 7).
These New York registry results argue for use of more restrictive criteria for PCI and more liberal referral of patients to CABG for revascularization, especially those with severe three-vessel disease. These data also demonstrate the importance of continued surveillance of outcomes when applying randomized trial data to the general patient population.
Cardiac valve surgery
Need for Valve Replacement
Asymptomatic aortic stenosis
A systolic velocity of ≥4 m/s by Doppler echocardiography identified 622 patients with isolated asymptomatic aortic stenosis who did not undergo surgery at the time of initial evaluation (29). This group, with an average age of 72 ± 11 years and 62% men, was followed for 5.4 ± 4.0 years. The probability of remaining free of symptoms and alive while unoperated was 82% at one year, 67% at two years, and 33% at five years (Fig. 8).Significant independent predictors of death on Cox modeling were advanced age (p < 0.0001), chronic renal failure (p < 0.004), inactivity (p < 0.001), and elevated aortic valve velocity (p < 0.03). Sudden death without preceding symptoms occurred in 11 (4.1%) of 270 unoperated patients. Because the risk of sudden death in this population is equivalent to that expected for aortic valve replacement, this study supports aggressive use of surgical management for asymptomatic patients with significant aortic stenosis.
Outcomes were evaluated in 456 patients prospectively defined to have asymptomatic organic mitral regurgitation, defined as a regurgitant volume of 66 ± 40 ml per beat and an effective regurgitant orifice of 44 ± 27 ml/m2, enrolled at a single institution between 1991 and 2000 (30). The population, with a mean age of 63 ± 14 years, 63% men, and EF mean of 70 ± 8, was followed for a mean of 2.7 ± 2.9 years under medical management (n = 224) or by medical management followed by surgery (mean interval 1.2 ± 2.0 years) in 203 patients followed 5.1 ± 2.9 years. Independent determinants of survival included advanced age, diabetes, and large effective regurgitant orifice. After adjustment for age, gender, diabetes, AF, and Charleson comorbidity index at baseline, the effective regurgitant orifice independently predicted survival with an adjusted risk ratio from any cause of 1.2 for each 10 ml/m2increment (p < 0.01). Regurgitant volume was not independently predictive of all-cause death. After multivariate analysis adjusted for age, gender, diabetes, AF, and effective regurgitant orifice, EF and Charleson comorbidity index, surgery decreased risk of death (hazard ratio 0.35, p < 0.01). Figure 9illustrates Kaplan-Meier estimates of rates of death from a cardiac cause among medically treated patients with asymptomatic mitral regurgitation stratified by effective regurgitant orifice that quantitate the maximal survival benefit possible with surgical intervention in asymptomatic patients as a function of duration of time after assessment.
Two cardiologists at a single center screened 682 consecutive patients between 1995 and 2002 with significant mitral regurgitation and systolic LV dysfunction on echocardiography who were identified to be candidates for mitral valve annuloplasty (MVA). After surgical evaluation, 126 patients underwent MVA (31). Propensity score derivation identified that age, EF, and LV dimension had been used to select patients for MVA. Mitral valve annuloplasty status treated as a time-dependent covariate was not an independent predictor of clinical outcome.
Prediction of operative risk
The Society of Cardiothoracic Surgeons of Great Britain and Ireland data from 16,679 patients who underwent heart valve surgery between 1995 and 2000 were used to develop a multivariate model to predict in-hospital mortality in patients undergoing mitral and/or aortic valve operations (32). This model performed well with a C index of 0.77 when validated in 16,160 other patients undergoing heart valve surgery from 2000 to 2003. A simplified risk score (to use with a lookup table) developed only from 2003 data provides a valuable bedside tool that permits clinicians to predict operative risks for individual patients undergoing aortic or mitral valve surgery over a range from mortality of 0.2% to more than 50% (Table 8).
Influence of race on risk
The Society of Thoracic Surgeons National Cardiac Database was used for retrospective review of 3,137 black and 46,249 white patients who underwent isolated mitral or aortic valve replacement from 1999 to 2002 (33). After adjustment for risk factors, black race was not a significant predictor of operative mortality but was associated with increased risk of morbidity.
Risks of reoperation
A single institution compared outcomes of 162 patients undergoing reoperative aortic valve replacement with 2,290 primary aortic valve replacements with or without CABG between 1993 and 2001 (34). Early mortality for reoperative aortic valve replacement was 5% compared to 3% (p = 0.20) for primary aortic valve replacement. Multivariate predictors for early mortality were prosthetic valve endocarditis, advanced preoperative functional class, peripheral vascular disease, and decreased LV EF. After adjustment by multivariable modeling, there was no difference in early mortality between the reoperative and primary aortic valve replacement groups. These data support the expanded use of bioprosthetic valves at a younger age because the mortality burden of repeat aortic valve replacement now approaches that of the primary operation.
Prosthetic Valve Failure
Published and unpublished data confirm that outlet structure fractures have occurred often with fatal outcomes in 633 (0.7%) of the 86,000 Björk-Shiley convexoconcave prosthetic heart valves implanted (35). Fractures continue to occur at a rate of 0.1% per year, so that risk of valve replacement now typically exceeds risk of sudden death associated with valve failure. Asymptomatic patients encountered with a Björk-Shiley prosthetic heart valve of unknown type should be carefully risk stratified using the implanted valve serial number to assess their individualized risk of acute valve failure for comparison with the risk of a 10% median overall in-hospital mortality for elective repeat mitral valve replacement.
Valve size match to patient size
A single-center study evaluated whether aortic valve prosthesis size relative to patient body size influenced functional recovery after aortic valve replacement using the Duke Activity Status Index completed before (n = 1,108) and after (n = 1,014) aortic valve replacement (36). Most patients improved functional capacity after operation, and neither indexed area (p = 0.94) nor standardized orifice size (p = 0.96) of the prosthetic valve inserted were predictors of functional recovery after aortic valve replacement.
Surgery for AF
The efficacy of left atrial radiofrequency ablation during mitral valve surgery in patients with chronic AF was assessed in a randomized double-blind trial at a single institution between December 2001 and November 2003 (37). A history of uninterrupted AF for six months was present in all 97 patients randomly assigned to undergo mitral valve surgery with (n = 48) or without (n = 49) ablation. Sinus rhythm was present one year after surgery in 20 (44.4%) of 45 ablated patients and in 2 (4.5%) of 44 patients with mitral valve surgery alone (p < 0.001). Secondary functional and activity capacity assessments showed improvement in the ablation cohort without an increase in postoperative mortality or morbidity.
Surgery for hypertrophic obstructive cardiomyopathy
The determinants of survival over a mean of 7.7 ± 5.2 years after surgical myectomy for obstructive hypertrophic cardiomyopathy were assessed in 338 adult patients (age 47 ± 14 years, 60% men) at a single institution between 1978 and 2002 (38). All patients had a maximal septal thickness of at least 13 mm and a preoperative LV outflow gradient of 66 ± 32 mm Hg. Operative deaths occurred in five (1.5%) of patients. Multivariable Cox regression analysis identified the five predictors of overall mortality to be advanced age, left atrial diameter ≥46 mm, female gender, concomitant CABG, and history of preoperative AF. Long-term survival was 98 ± 1% at one year, 95 ± 1% at 5 years, and 83 ± 3% at 10 years.
Comparison of myectomy and alcohol ablation
A single-center retrospective study comparing septal myectomy to alcohol ablation was conducted in 150 symptomatic adults with hypertrophic obstructive cardiomyopathy between November 1998 and December 2003 (39). Surgical myectomy was performed in 90 patients referred directly for the procedure and in five other patients after failed attempted alcohol ablation. A concomitant procedure was performed in 47 of these patients. The 54 patients who received alcohol ablation and the 48 surgical patients who had an isolated myectomy had a mean follow-up of 2.2 years after the procedure. There were five late deaths after alcohol ablation and one after myectomy. No significant difference was observed in after-intervention septal thickness or freedom from after-intervention pacing. However, three-year freedom from pacing was 88% in the surgically treated patients compared with 59% in the ablation-treated patients. Until larger studies are available, surgical myectomy should be considered the standard to which alcohol ablation must be compared.
Surgery for congenital heart disease
Cardiopulmonary Bypass in Children
Fresh compared to reconstituted blood for prime
A single-center randomized, double-blind controlled trial of cardiopulmonary bypass circuit priming with fresh (96 patients) or reconstituted (104 patients) blood was conducted in children <1 year of age (40). The group that received reconstituted blood had the shorter stay in the intensive care unit (70.5 h vs. 97 h, p = 0.04) and a smaller cumulative fluid balance at 48 h. Bleeding appeared similar in both groups. The use of fresh whole blood for cardiopulmonary bypass in children provides no advantage over a pump prime using packed red blood cells and fresh-frozen plasma.
Management of children with C4A deficiency
A single study was conducted in 116 children lacking complement component C4A and susceptible to capillary leak syndrome during cardiopulmonary bypass (41). Three (5%) of 58 patients given C4A-rich plasma priming of the bypass circuit had capillary leak syndrome, compared with 56 (97%) of patients given C4A free plasma (p < 0.0001). A high concentration of inflammatory markers occurred postoperatively in patients with capillary leak syndrome. C4A-rich plasma priming appears to reduce the incidence of capillary leak syndrome after cardiopulmonary bypass by markedly preventing the increase in activated C4 and by attenuating the increase in other inflammatory mediators.
Outcomes of Surgery
Influence of case volume
Among the 29 pediatric hospitals of the Child Health Corporation of America, 87 surgeons performed 801 Norwood procedures for hypoplastic left heart syndrome from January 1998 to December 2001 (42). Total institutional Norwood operation volumes averaged 5.15 (range 2 to 109). Survival ranged from 25% to 100% among institutions. The average number of Norwood operations per surgeon was 10.7 ± 14 over the four-year period. The overall operative survival rate to hospital discharge was 68%. Stepwise linear regression showed hospital case volume to relate to survival (p = 0.02), but surgeon case volume did not.
Intracoronary ultrasound late after arterial switch operation
The status of reimplanted coronary arteries was evaluated by cardiac catheterization and intravascular ultrasound performed in 22 asymptomatic children (median age 9.5 years) after arterial switch operation (interval range 5 to 12.6 years) (43). Coronary angiography identified a single significant stenosis of the right coronary ostium. Quantitative ultrasound analysis of 37 coronary artery segments in 20 children showed some degree of intimal proliferation in 33 of 37 coronary arteries (89%). Significant intimal proliferation (>0.3 mm) was observed in 10 of the 20 patients. Signs of development of early atherosclerosis in reimplanted coronary arteries will require future monitoring to assess disease progression and the likelihood of occurrence of late coronary events.
Status report of major ongoing randomized clinical trials in cardiac surgical patients
Outcomes Following Myocardial Revascularization: On and Off Cardiopulmonary Bypass (ROOBY Trial)
Sponsor:Veterans Affairs Cooperative Study.
Principal Investigator:Frederick L. Grover, MD ().
Hypothesis:In properly selected patients, 30-day and one-year mortality and morbidity are equivalent when CABG is conducted with or without cardiopulmonary bypass.
Enrollment:1,763 of 2,200 patients enrolled by 17 U.S. Veterans Affairs hospitals.
Surgical Treatment for Ischemic Heart Failure (STICHTrial)
Sponsor:National Institutes of Health.
Principal Investigator:Robert H. Jones, MD ().
Hypothesis 1:Coronary artery bypass grafting combined with intensive medical therapy improves long-term survival compared to medical therapy alone in 1,000 patients with LV EF ≤0.35 and CAD amenable to CABG.
Hypothesis 1 Enrollment:636 of 1,200 patients enrolled by 83 sites worldwide; completion expected December 2006.
Hypothesis 2:Adding surgical ventricular restoration to CABG and medical therapy improves survival free of cardiac hospitalization in 1,000 patients with LVEF ≤0.35, CAD amenable to CABG, and dominant anterior akinesia or dyskinesia.
Hypothesis 2 Enrollment:1,000 patients, completed January 24, 2006, at 96 sites worldwide.
Future Revascularization Evaluation in Patients With Diabetes Mellitus: Optimal Management of Multivessel Disease (FREEDOM) Trial
Sponsor:National Heart, Lung, and Blood Institute.
Principal Investigator:Valentin Fuster, MD, PhD ().
Hypothesis:In 2,400 diabetic patients with multi-vessel disease amenable to CABG and PCI and indication for revascularization, either CABG or PCI will be superior in preventing death, nonfatal MI, and stroke.
Enrollment:60 of 2,400 patients enrolled by 9 clinical sites.
Dr. Jones acknowledges Vanessa Moore with appreciation for her skill in organization and presentation of information that is apparent in the tables and her intense motivation that permitted this manuscript to be completed in a brief interval after the close of 2005.
↵1 Dr. Jones is the principal investigator of the National Institutes of Health-sponsored Surgical Treatment for Ischemic Heart (STICH) Failure Trial
- Received January 20, 2006.
- Revision received January 31, 2006.
- Accepted February 7, 2006.
- American College of Cardiology Foundation
- Liao L.,
- Kong D.F.,
- Shaw L.K.,
- et al.
- Zacharias A.,
- Schwann T.A.,
- Riordan C.J.,
- Durham S.J.,
- Shah A.,
- Habib R.H.
- Konety S.H.,
- Vaughan-Sarrazin M.S.,
- Rosenthal G.E.
- Chareonthaitawee P.,
- Gersh B.J.,
- Araoz P.A.,
- Gibbons R.J.
- Kapetanakis E.I.,
- Medlam D.A.,
- Boyce S.W.,
- et al.
- Okrainec K.,
- Platt R.,
- Pilore L.,
- Eisenberg M.J.
- Cheng D.C.,
- Bainbridge D.,
- Martin J.E.,
- Novick R.J.,
- Evidence-based Perioperative Clinical Outcomes Research Group
- Legare J.F.,
- Buth K.J.,
- Hirsh G.M.
- Wijeysundera D.N.,
- Beattie W.S.,
- Djaiani G.,
- et al.
- Mercado N.,
- Wijns W.,
- Serruys P.W.,
- et al.
- Surruys P.W.,
- Ong A.T.L.,
- van Herwerden L.A.,
- et al.
- Pellika P.A.,
- Sarano M.E.,
- Nishimura R.A.,
- et al.
- Wu A.H.,
- Aaronson K.D.,
- Bolling S.F.,
- Pagani F.D.,
- Welch K.,
- Koelling T.M.
- Ambler G.,
- Omar R.Z.,
- Royston P.,
- Kinsman R.,
- Keogh B.E.,
- Taylor K.M.
- Taylor N.E.,
- O’Brien S.O.,
- Edwards F.H.,
- Peterson E.D.,
- Bridges C.R.
- Blot W.J.,
- Ibrahim M.A.,
- Ivey T.D.,
- et al.
- Koch C.G.,
- Khandwala F.,
- Estafanous F.G.,
- Loop F.D.,
- Blackstone E.H.
- Woo A.,
- Williams W.G.,
- Choi R.,
- et al.
- Pedra S.R.F.F.,
- Pedra C.A.C.,
- Abizaid A.A.,
- Braga S.L.N.,
- et al.