Author + information
- Robert H Jones, MD, FACC* ()
- ↵*Reprint requests and correspondence:
Dr. Robert H. Jones, P.O. Box 17969, Duke Clinical Research Institute, Durham, North Carolina 27715, USA.
This article highlights information from cardiovascular surgery articles published in 2003 that are likely to have a broad impact on the practice of cardiovascular specialists in 2004. A scan of major general and cardiovascular journals, complemented by a formal MEDLINE search, focused on identifying well-conducted randomized or observational studies of clinical outcomes that should gide cardiovascular surgery care management decisions for individual patients with specific cardiovascular disease presentations. Preference has been given to reporting articles that are fresh in vision, thoughtful in design, and definitive in conclusions.
Quality of cardiovascular surgical care
Influence of procedure volume on mortality
Table 1summarizes data from three separate studies that revisit the question of whether surgeon and hospital coronary artery bypass grafting (CABG) case volumes remain independent predictors of mortality (1–3). All three large data sets that used different specific thresholds to define low and high volumes of CABG by surgeons and hospitals suggest that patients treated by low-volume surgeons at low-volume hospitals may sustain as much as an additional 1.9% mortality risk compared with those treated by high-volume surgeons at high-volume institutions. Analysis of the New York State (NYS) data suggested that 125 cases per surgeons to be the optimal threshold of surgeon volume to maintain quality without undue disruption of access to care. For example, moving the 17,066 NYS patients from low- to high-volume surgeons and hospitals would be expected to prevent 106 patient deaths and decrease overall operative mortality by 0.6%. This analysis suggests a potential regulatory approach to improving the outcomes of care of CABG patients. However, a more appealing approach to ensuring that patients receive the highest quality of CABG care would be for all cardiac surgeons to voluntarily monitor their individual hospital risk-adjusted mortality statistics, using the Society of Thoracic Surgeons (STS) database.
Figure 1depicts the relationship between risk-adjusted CABG mortality and annual case volume for the 439 hospitals now using this approach to monitor CABG outcomes. Annual hospital CABG volumes ranged from 39 to 1,754 (median 253, interquartile range 165 to 417), and only 18% of STS centers performed more than 500 isolated CABG procedures per year. An association between hospital volume and outcome persisted after adjusting for patient risk and clustering effects (p = 0.004). Moreover, the widest variance was observed among low-volume centers. Therefore, it is especially important for cardiac surgeons who operate at low- or medium-volume hospitals to prospectively monitor and report risk-adjusted mortality to referring cardiologists and patients, using standard methodology.
Other observations on the relationship between case volume and patient outcome
Low-risk patients benefit as much or more than high-risk patients by receiving care at high-volume hospitals (4).
A study of MEDPAR data from 684 hospitals that performed more than 1.5 million heart operations between 1994 and 1999 showed CABG hospital mortality rates correlated closely with mortality rates for valve surgery (5).
Provider profiling to improve CABG quality
Generation and publication of individual hospital and surgeon CABG mortality rates, as a tool for public education and to gide quality improvement initiatives, have been praised by some for apparently lowering mortality rates and vilified by others for discouraging surgeons from appropriately intervening in high-risk patients. To examine the influence of systematic monitoring of quality on outcomes of cardiac surgery, administrative data in MEDPAR-Part A were analyzed for 911,407 ≥65-year-old patients undergoing CABG in the U.S. during the 1994 to 1999 period (6). A risk adjustment model derived from administrative data functioned with a cstatistic of 0.71 in predicting the risk of death and actual 30-day in-hospital mortality rates. Except for New Jersey, all five regions with prospective CABG mortality monitoring programs had lower CABG mortality than the remainder of the U.S. For example, risk-adjusted death in the hospital or within 30 days was 3.15 in 52,209 NYS patients, compared with 4.75 in 756,132 patients in U.S. regions without prospective mortality monitoring. There was no evidence of migration of patients out of NYS for CABG during this interval, which might suggest that this program intimidated surgeons from accepting high-risk patients. Over this entire time interval, the adjusted odds ratio for all five regions with prospective CABG mortality monitoring was 0.7 of the CABG mortality observed in all other regions of the country. This magnitude of risk reduction suggests that implementation of a national CABG reporting program would prevent 1,800 needless CABG deaths annually in this country.
Continuous quality improvement intervention improves CABG processes of care
Using the STS national cardiac data base, 359 academic and nonacademic hospitals treating 267,917 patients undergoing CABG between January 2000 and July 2002 were randomized into a control arm or one of two groups that used continuous quality improvement interventions to increase the preoperative use of beta-blockade and the incidence of internal mammary artery (IMA) grafting in elderly patients (7). After obtaining baseline measurements at all sites, a sequence of three intervention rounds was begun in the two cohorts randomized to increased beta-blocker or IMA graft use by distribution of materials to a local champion at these sites. Centers in the control arm received no interventional material. All groups showed an increase in the two process measures over time, but beta-blocker use was 7.3% more likely at sites receiving focused intervention on its appropriate preoperative use. Grafting of the IMA increased 8.7% more in elderly patients at IMA intervention sites, compared with control sites. Although the magnitude of process improvement appeared modest, the use of professional society resources to voluntarily conduct quality improvement initiatives provides an important model worthy of adoption by other professional organizations.
Surgery for coronary artery disease (CAD)
Comparison of CABG versus percutaneous coronary intervention (PCI)
A meta-analysis of events occurring up to eight years after randomization of 7,964 patients into 13 clinical trials compared outcomes of PCI and CABG performed during the 1987 to 1999 period (8). For all patients entered in all 13 trials, a 2% absolute survival benefit favoring CABG was observed at five years but not at earlier or later time intervals (Fig. 2). Trials enrolling only patients with multivessel or proximal left anterior descending coronary artery (LAD) single-vessel CAD showed a significant survival advantage of CABG over PCI-treated patients at five years, which at eight years had increased to a 3.4% absolute survival advantage. Patients with single-vessel proximal LAD disease experienced a 5.6% risk reduction at five years for CABG compared with PCI. Data from four trials reporting subsets of diabetic patients showed CABG provided an 8.6% survival advantage compared with PCI at four years, but this difference became insignificant by 6.5 years. Repeat revascularization rates were lower with CABG compared with PCI, although stents in more recent trials appeared to reduce the magnitude of this difference. In summary, these data show a modest survival advantage of CABG over PCI in patients with multivessel disease. Longer term follow-up of more recent trials that treated patients with modern stenting may lessen this advantage. This meta-analysis of currently available data provides a framework for assisting patients to select the most desirable form of revascularization for their specific clinical situation.
Recent CABG/PCI trial
A randomized comparison of 280 patients referred for PCI showed stenting to be more cost effective than off-pump surgery at one year, while maintaining comparable cardiac outcomes and quality of life (9).
On- and off-pump CABG comparison
A meta-analysis of six studies, including 558 patients randomized to CABG performed on cardiopulmonary bypass and 532 randomized to CABG without bypass, found no significant difference in the combined end point of death, stroke, or myocardial infarction (10).
Randomized comparison of primary CABG using the radial artery or right internal thoracic artery or saphenous vein as a randomly assigned conduit
A prospective, randomized, single-center trial addressed the hypotheses that: 1) angiographic patency of the radial artery is superior to that of the free right internal thoracic artery or saphenous vein graft; and 2) patients who receive radial artery grafts would have superior outcomes to those who receive free right internal thoracic artery or saphenous vein grafts (11). All patients received an in situ left internal thoracic artery graft to the LAD. The second graft utilized was assigned by stratified randomization to be either a radial artery or right internal thoracic artery in young patients or the radial artery or saphenous vein in older patients. Saphenous veins were used for subsequent grafts, as needed. Between June 1996 and February 2002, 438 patients were enrolled: 285 assigned to the younger group and 153 assigned to the older group. No significant difference was seen in survival or cardiac event-free survival. Angiography was performed by random assignment at predetermined intervals between 0 and 10 years. Five-year angiography was offered to all patients. The five-year patency rates were 95% for radial arteries and 100% for right internal thoracic arteries in group 1. The five-year patency rates were 87% for radial arteries and 94% for saphenous veins in group 2. The five-year clinical and angiographic results of this study do not support the hypothesis of greater patency or freedom from untoward events associated with radial artery conduit use.
Conduit patency rates
The University of Melbourne group assessed patency in 1,402 CABG patients on 3,218 saphenous vein graft angiograms obtained between 1977 and 1999 at a mean interval from CABG to repeat angiogram of 9.3 years and used multivariate analysis to identify factors independently influencing graft patency (12). Grafts with <80% stenosis were considered patent. Table 2summarizes patient and operative variables found to be signifi-cantly related to conduit patency at the level of ≤0.002. Even after correction for the influence of the interval from operation to angiography, the year of operation remained a predictor of occlusion. The fact that grafts placed in 1990 would be expected to have a 0.37 rate of occlusion, compared with grafts performed in 1980, over comparable intervals to angiography, illustrates the fallacy of comparing old saphenous vein patency information to more recently evaluated arterial conduits. Younger age predicts worse graft patency, as other causes of death remove older patients from the study population at a greater rate, before graft occlusion becomes apparent. Greater blood flow through conduits leading to the LAD or to large arteries results in lower conduit occlusion. Because arterial grafts are commonly used preferentially for bypassing the LAD and other coronary arteries with high blood flow, confounding can always be expected in comparisons of arterial and venous grafts. Lower flow velocity in saphenous vein conduits with larger diameters appears to be the factor leading to higher occlusion of these suboptimal vessels. Knowledge of these factors leads to the proper interpretation of modern reports on conduit patency from small groups of patients followed for brief intervals.
The University of Melbourne group also prospectively assessed conduit patency in 1,408 symptomatic patients after CABG (13). Data tabulated for each type of conduit evaluated over time reflect surprisingly high patency rates, even in these symptomatic patients (Table 3). This information is useful to address patient issues regarding optimal conduit use in discussion before the anticipated CABG operation.
Shock and CABG
A consecutive series of 259 patients undergoing CABG while in cardiogenic shock or receiving cardiopulmonary resuscitation between 1979 and 2001 were studied using time-related and multivariate methodologies (14). Survival was 59% at 1 year and 47% at 10 years. A substantial early hazard was followed by a normal hazard of late death. On multivariate analysis, comorbidity, treated diabetes, a lower pH at initiation of the operation, and the presence of three-vessel disease were independent predictors of lower survival. Advanced age contributed little, but most of the population was young. The discriminating power for prediction of hospital mortality using combinations of these variables was low.
Data from the SHould we emergently revascularize Occluded Coronaries for cardiogenic shocK? (SHOCK) trial registry now raise questions about the applicability of previously reported results in the prespecified ≥75-year-old patient subgroup of the randomized cohort, which suggested that CABG had no survival benefit (15)(Fig. 3). During the time these 56 patients were being randomized into the SHOCK trial, 277 patients (age ≥75 years) were enrolled into the SHOCK registry. According to the SHOCK registry, elderly patients who underwent early revascularization had a lower covariate-adjusted mortality than those who had revascularization late or never. Characteristics of elderly patients that suggested a better outcome with revascularization may have led physicians to select the registry in preference to randomization. This registry finding raises the question of the generalizability of the lack of benefit observed in the randomized cohort. Advanced age should not be a categorical reason to deny CABG to patients in shock.
Combined carotid artery endarterectomy and CABG
A review of 10,561 carotid endarterectomy reports from Medicare patients in 10 states identified 226 procedures performed in combination with CABG (16). Complete medical records also were reviewed in 184 of these patients. A recent stroke or transient ischemic attack was the indication for carotid endarterectomy in only 12% of these patients, and 56% of these patients were not symptomatic from their carotid disease. Procedure-related mortality was 6.6%, and the combined end point of stroke and death was 17.7%. Moreover, 80% of nonfatal strokes were disabling. These community-based results of combined CABG and carotid endarterectomy, which reflect a 15% incidence of death or disability, suggest that use of combined carotid endarterectomy and CABG is more widespread than is justified, especially in this population with a high representation of asymptomatic carotid disease.
Surgery for valvular heart disease
Mitral regurgitation (MR) management in patients undergoing CABG
The management of mild to moderate MR in patients undergoing CABG remains controversial. Increasingly, outcomes are being reported on patients grouped by the operation done and not by an objective assessment of MR severity. Because mitral valve operations that might be considered questionably indicated occur in patients with less severe MR, the decision to perform a mitral valve operation immediately shifts one of the sickest of the CABG-only patient cohorts to a position as one of the least sickest members of the CABG/mitral surgery cohort. Therefore, a low threshold for performing mitral surgery during CABG improves mortality in both the CABG-only and CABG/mitral surgery subgroups if the subgroups are defined by operation performed and not by a measurement of disease severity.
A prospective study of 1,939 consecutive CABG patients included 167 patients in whom preoperative and intraoperative echocardiograms objectively defined the presence of mild to moderate MR, for whom some surgeons might add mitral valve surgery to CABG but for whom CABG without mitral valve surgery was performed (17). These patients with MR had an observed 11% decrement in 5-year survival and an 18% decrement in 10-year survival, compared with patients without MR. However, after adjusting survival by multivariate Cox regression analysis using advanced age, previous stroke, poor left ventricular function, heart failure, diabetes, peripheral vascular disease, and no left internal thoracic artery use, mild to moderate MR was not an independent predictor of long-term mortality. Because survival differences observed in the MR subgroup were fully explained by other variables, operative repair of MR would have been unlikely to change survival. This long-term report of the follow-up of CABG patients presenting with mild to moderate MR and treated with revascularization alone provides an important framework to design prospective, randomized studies to definitively address this important issue.
Prosthetic valve performance
A prospective study compared outcomes of the Starr Edwards and St. Jude prostheses inserted in the aortic (267 patients) and mitral positions (122 patients) (18). Patients were randomized during the operative procedure after the valve was excised and the operating surgeon was at equipoise with the value of the two prostheses. The specific valve was allocated by minimization using a computer program to stratify important clinical characteristics. Actuarial event-free survival did not differ for either the mitral or aortic valve comparisons. Symptom relief was similar at five years after surgery, and no significant difference was found in complication rates during follow-up, except that eight patients with Starr Edwards aortic valves developed infectious endocarditis, compared with only one patient with a St. Jude aortic valve. Left ventricular function had a marked effect on long-term survival that overwhelmed the small, insignificant differences that might be related to the different prostheses used. This study suggests that many of the touted benefits of specific cardiac valves discussed in observational studies might be of less significance or absent if cardiac valves were subjected to the rigor of randomized clinical trials.
Prognosis in endocarditis
A retrospective observational cohort study conducted from January 1990 to January 2000 at seven Connecticut hospitals derived a multivariate model to predict death in 259 adults with left-sided native valve endocarditis, which was prospectively validated in a separate cohort of 254 patients (19). The six-month mortality rate used as the modeling end point was 25% in the derivation and 26% in the validation samples. Five baseline features were independently associated with six-month mortality and, after validation, were used to derive a simple scoring system to predict mortality for individual patients (Table 4). The fact that underuse of surgery appeared to contribute independently to mortality suggests that a lower threshold for operation used prospectively would be likely to decrease mortality. Until further studies examine the interaction between use of surgery and other prognostic variables to refine information that might suggest the value of surgery is confined to specific subgroups, it is reasonable to consider patients increasingly as surgical candidates as a function of their predicted mortality.
Optimal anticoagulation with mechanical heart valves
A meta-analysis from 35 studies, including 23,145 total patients studied for 108,792 patient-years, suggests that the optimal intensity of vitamin K antagonists in patients with mechanical heart valves should range from an international normalized ratio of 3.0 to 4.5, with patients with aortic valves managed at the lower side of this range and patients with mitral valves managed at the higher side of this range (20).
Surgery for congenital heart disease
Neuroprotection during cardiopulmonary bypass in infants
Neurodevelopment status was evaluated in 155 infants with dextro-transposition of the great arteries, who by random assignment underwent either total circulatory arrest or low-flow cardiopulmonary bypass for neuroprotection during an arterial switch operation (21). Treatment groups did not differ on most subsequent determinations of neurologic status, including IQ score, academic achievement, memory, problem solving, and visual motor integration. However, children assigned to total circulatory arrest performed worse on tested motor function. Although the mean scores on most outcomes were within normal limits, neurodevelopmental status in the cohort as a whole was below expectations in many dimensions. In a companion paper, nonparametric regression and piecewise linear models indicated that neurodevelopmental outcomes were generally not adversely affected, unless the duration of circulatory arrest exceeded a threshold of 41 min (22). However, this threshold was not sufficiently discriminatory to be interpreted as a universally safe duration for total circulatory arrest.
Influence of hemodilution on outcome of cardiopulmonary bypass in infants
A single-center, randomized trial using two hemodilution protocols during hypothermic cardiopulmonary bypass was conducted in 174 infants undergoing cardiac surgery (23). The cardiopulmonary bypass pump was configured to attain a hematocrit of 21.5 in the cohort randomized to a low hematocrit and to 27.8 in the cohort randomized to a high hematocrit. At age one year, the lower hematocrit group had worse scores on the Psychomotor Development Index (81.9 ± 15.7 vs. 89.7 ± 14.7, p = 0.008), demonstrating the practice of hemodilution to a low hematocrit level currently considered safe to be associated with adverse developmental outcomes.
Pharmaceuticals for support of children undergoing heart surgery
A placebo-controlled trial randomized pediatric congenital heart surgery patients to placebo, low-dose milrinone, or high-dose milrinone and evaluated the composite end point of death or development of low cardiac output syndrome at 36 h and 30 days (24). Among 238 treated patients, 25.9%, 17.5%, and 11.7% in the placebo, low-dose milrinone, and high-dose milrinone groups, respectively, developed low cardiac output syndrome in the first 36 h after the operation. High-dose milrinone significantly reduced the risk of development of low cardiac output syndrome, as compared with placebo (p = 0.023). This trial provides evidence suggestive of a benefit in a patient group that has not been extensively studied. However, the results must be considered only suggestive, because clinicians who were blinded as to the drug could easily recognize children receiving placebo infusion by their blood pressure stability, in contrast to the acute hypotension consistently induced by milrinone infusion. Moreover, only two deaths occurred, and both were associated with infusion of the study drug. Therefore, the trial only showed a change in the clinical diagnosis of low cardiac output syndrome and not in actual patient outcomes.
A double-blinded, randomized, placebo-controlled study evaluated the efficacy of aprotinin in attenuating hemostatic and inflammatory activation during cardiopulmonary bypass in 60 patients weighing <10 kg (25). Secondary end points included the influence of aprotinin on reduction of blood loss, blood transfusion requirement, adequacy of postoperative oxygenation, and length of mechanical ventilation. The 24-h chest tube drainage was 5.9 ml/kg less during the first 24 h after the operation in aprotinin-treated children. Blood transfusion beyond that used to prime the heart–lung machine for all operations was required in 13% of aprotinin-treated and 47% of placebo-administered children (p < 0.05). Also, postoperative oxygenation was lower and time on the ventilator was shorter in the aprotinin group. No side effects were attributed to aprotinin. High-dose aprotinin may be effective in attenuating hemostatic activation in reducing blood loss in pediatric cardiac surgery.
Pulmonary embolism in adult Fontan patients
A group of 30 adult Fontan outpatients without specific symptoms underwent ventilation-perfusion scans to detect pulmonary embolism (26). A positive scan, which confirmed pulmonary embolism by pulmonary angiography, was present in 5 (17%) of the 30 patients. No pulmonary emboli were present in the nine patients taking warfarin. A prospective trial evaluating the importance of anticoagulation therapy in adult Fontan patients appears warranted.
Status report of major ongoing randomized clinical trials in cardiac surgical patients
Outcomes after myocardial revascularization: on and off cardiopulmonary bypass
Veterans Affairs Cooperative Study.
Frederick L. Grover, MD (>).
In properly selected patients, 30-day and 1-year mortality and morbidity are equivalent when CABG is conducted with or without using cardiopulmonary bypass.
Secondary end points
Angiographic graft status, completeness of revascularization, recurrent ischemia, neuropsychological status, quality of life, and resource use.
Randomization stratified by surgeon; urgent or elective isolated CABG procedures; 2,200 patients randomized within four years; one-year follow-up.
February 2004 status
Total of 3,214 patients screened, 858 randomized; major exclusion: small, diffusely diseased coronary arteries (35%); conversion rates: 9.9% for off- to on-pump and 2.5% for on- to off-pump; outcomes of total group: death in 3.2%, perioperative myocardial infarction 0.3%, return to the operating room for bleeding (48 h) 1.8%, cardiac arrest 1.6%, repeat cardiac surgery 0.2%, stroke 1.1%, sternal wound infection 1.4%, and renal failure 1.3%.
Seventeen U.S. Veterans Affairs hospitals.
Surgical Treatment for IschemiC Heart failure (STICH trial)
National Institutes of Health, Bethesda, Maryland.
Robert H. Jones, MD ().
1) CABG combined with intensive medical therapy (MED) improves long-term survival, compared with MED alone, in 2,000 patients with a left ventricular ejection fraction ≤35% and CAD amenable to CABG; 2) surgical ventricular restoration (SVR) combined with CABG and MED improves survival free of cardiac hospitalization, compared with CABG and MED without SVR.
Secondary end points
Cardiac mortality and morbidity, resource use, quality of life, noninvasive cardiac study prediction of primary end points by myocardial ischemia and viability, left ventricular size and function, neurohormonal and cytokine levels, and genetic polymorphisms or mutations.
A total of 2,800 patients with a left ventricular ejection fraction ≤35% and CAD amenable to CABG will be stratified on the basis of MED eligibility (no left main stenosis ≥50% or angina class ≥III by the Canadian Cardiovascular Society) and SVR eligibility (dominant anterior akinesia or dyskinesia amenable to SVR). The allocation of patients to treatment within each stratum is illustrated in Figure 4. Patients eligible for either MED or CABG but not eligible for the SVR procedure (stratum A) will be randomized in equal proportions to MED or CABG. Patients eligible for all three therapies (stratum B) will be randomized in equal proportions to MED, CABG, or CABG plus SVR. Patients whose angina or CAD severity makes them ineligible for MED but who are appropriate candidates for both CABG and SVR (stratum C) will be randomized in equal proportions to CABG or CABG plus SVR. Use of patients in stratum B assigned to CABG, to address both primary hypotheses, provides 3,000 end points from 2,800 enrolled patients.
February 2004 status
Table 5demonstrates the characteristics of 261 randomized patients.
North American, European, Asian-Pacific, and South American sites added at a rate of five per month until the goal of 100 enrolling sites was achieved. Enrollment projected for completion is December 2006 for the revascularization hypothesis and April 2005 for the SVR hypothesis.
- coronary artery bypass grafting
- coronary artery disease
- internal mammary artery
- left anterior descending coronary artery
- medical therapy
- mitral regurgitation
- New York State
- percutaneous coronary intervention
- Society of Thoracic Surgeons
- surgical ventricular restoration
- American College of Cardiology Foundation
- Hannan E.L,
- Wu C,
- Ryan T.J,
- et al.
- Hoffman S.N,
- TenBrook J.A,
- Wolf M.P,
- Pauker S.G,
- Salem D.N,
- Wong J.B
- Eefting F,
- Nathoe H,
- van Dijk D,
- et al.
- Sergeant P,
- Meyns B,
- Wouters P,
- Demeyere R,
- Lauwers P
- Dzavik V,
- Sleeper L.A,
- Cocke T.P,
- et al.
- Vink R,
- Kraaijenhagen R.A,
- Hutten B.A,
- et al.
- Hoffman T.M,
- Wernovsky G,
- Atz A.M,
- et al.
- Varma C,
- Warr M.R,
- Hendler A.L,
- Paul N.S,
- Webb G.S,
- Therrien J