CABG Survival Beats PCI in Big U.S. Dataset

CHICAGO – Coronary artery bypass surgery had a significant 4-year survival advantage of 4 percentage points over percutaneous coronary intervention, based on analysis of an observational, U.S.-wide dataset with nearly 190,000 patients.

These results, from a new collaboration between the American College of Cardiology Foundation (ACCF) and the Society for Thoracic Surgeons (STS), confirmed findings made by other, recent randomized and observational studies that coronary artery bypass grafting produces a modest, long-term survival advantage in stable patients undergoing revascularization, compared with percutaneous coronary intervention. The new finding seemed notable because it came from an unprecedentedly large U.S. dataset that broadly represented U.S. practice in the mid 2000s, and because the analysis used sophisticated statistical corrections and tests to try to eliminate patient-selection factors as confounders.

ASCERT, the ACCF and STS Database Collaborative on the Comparative Effectiveness of Revascularization Strategies, "is larger than any previous study [comparing coronary artery bypass surgery and percutaneous coronary intervention], it’s contemporary data [from cases treated during 2004-2007], and it’s more generalizable because it covers the entire United States," although limited exclusively to Medicare patients who were at least 65 years old, said Dr. William S. Weintraub, lead investigator of the study, at the annual meeting of the American College of Cardiology.

But several cardiologists who heard the report cautioned that despite all the analytic efforts by Dr. Weintraub and his associates to craft a comparison that eliminated biases, the results had limited implications because the analysis had an inherent inability to provide a truly clean comparison of coronary artery bypass surgery (CABG) and percutaneous coronary intervention (PCI) due to its reliance on observational data.

"It’s fair to say that at least in this data set, it’s impossible to eliminate residual confounders or selection bias because of the factors that a cardiologist or surgeon integrate at the bedside, such as frailty, patient preference, bleeding risk, and compliance with medical therapy," commented Dr. Alice K. Jacobs, a cardiologist and professor of medicine at Boston University. "I think these results will probably not change clinical practice."

"You can do multivariate, propensity analyses till kingdom come, but it will never eliminate the selection biases," in these data, said Dr. Bernard J. Gersh, a cardiologist and professor of medicine at the Mayo Clinic in Rochester, Minn.

"A major flaw in ASCERT was that it compared apples to oranges. It looked at a specific subset of very sick, older patients, and it did not take into account a variety of factors that could have affected these data," said Dr. Christopher J. White, president of the Society for Cardiovascular Angiography and Intervention in a written statement.

"PCI is not only a safe and effective treatment option for patients, it often is the only treatment available to patients too sick or too frail to have open heart surgery," added Dr. White, who is also professor and chairman for cardiovascular diseases at the Ochsner Heart and Vascular Institute in New Orleans.

Despite the limitations of the study, it may result in a small but meaningful resetting of how coronary revascularization is used, Dr. Weintraub said. "Practice has shifted toward PCI; this may change it back a little bit. But I don’t think it will have a tremendous effect by itself, and it probably shouldn’t," said Dr. Weintraub, chief of cardiology at Christina Care Health System in Newark, Del. "ASCERT largely supports the [coronary revascularization] guidelines that are already there," issued last year by the ACC, the American Association for Thoracic Surgery, and other organizations, he added (J. Am. Coll. Cardiol. 2011;58:2584-614; J. Am. Coll. Cardiol. 2011;58:2550-83).

"I hope we don’t have people saying that this shows that all patients should go to surgery," said cardiac surgeon Dr. Fred H. Edwards in an interview. "This is one more piece of information to use when evaluating how to treat patients with stable coronary artery disease. The information should be presented to patients and used in a dialog between surgeons and cardiologists," said Dr. Edwards, professor emeritus of surgery at the University of Florida in Jacksonville, director of the STS Research Center in Chicago, and coprincipal investigator for ASCERT.

"The key to ASCERT is its generalizability. In ASCERT we showed, hopefully definitively, that long-term survival in patients with stable two- and three-vessel coronary disease is better with surgery. But is survival the only thing that’s important? No, there is also the stroke rate, long-term quality of life, and patient preference."

The current ASCERT analysis focused on survival data for 86,244 patients who underwent CABG – data collected in the STS Adult Cardiac Surgery Database – and 103,549 patients who underwent PCI from the ACC Foundation’s National Cardiovascular Data Registry. The study’s central finding was that after adjustment by inverse probability weighing the 4-year mortality rate among CABG patients was 16.4%, and 20.8% among PCI patients. A second key finding was that initially after intervention, mortality was lower with PCI. The survival curves of the two treatment groups did not cross until 1 year out from the procedure, after which the survival advantage following CABG gradually increased over time.

Future reports will focus on other outcomes, including stroke rates, Dr. Weintraub said. Concurrent with his report at the meeting an article on the findings appeared online (N. Engl. J. Med. 2012 March 27 [doi:10.1056/NEJMoa1110717]).

Perhaps the most striking element in the findings was that the long-term survival benefit with CABG over PCI was consistent across a variety of subgroup analysis, including both women and men, and patients with or without diabetes.

"The advantage of CABG in all subgroups was a major surprise," Dr. Edwards said in an interview. "We thought that we’d see some subgroups that would benefit from PCI. Much to our surprise, all the subsets showed better survival with surgery, generally in the range of 20%-30%" on a relative basis.

Dr. Weintraub highlighted some extra analytic steps he and his associates took in the study. In addition to the primary adjusted analysis, the researchers performed a second, propensity-matched analysis on a subgroup of about 43,000 CABG patients and an equal number of PCI patients who closely matched for a list of over 20 demographic and clinical variables. The propensity-matched calculations found a similar long-term survival advantage for CABG.

They also examined whether the observed differences could be explained by an unmeasured confounder, such as frailty. To explain the observed between-group difference, "the unmeasured confounder would have to have an effect so large that it would have a hazard ratio of two, and would need to occur in about 30% of the patients in one group and in only 10% of those in the other group. How could we have missed that?" Dr. Edwards said.

Dr. Weintraub, Dr. Edwards, and Dr. White said they had no relevant disclosures. Dr. Jacobs said she had received research grants from Abiomed, Accumetrics, and Abbott Vascular. Dr. Gersh said he has been a consultant to Boston Scientific and Abbott Laboratories.

CHICAGO – Coronary artery bypass surgery had a significant 4-year survival advantage of 4 percentage points over percutaneous coronary intervention, based on analysis of an observational, U.S.-wide dataset with nearly 190,000 patients.

These results, from a new collaboration between the American College of Cardiology Foundation (ACCF) and the Society for Thoracic Surgeons (STS), confirmed findings made by other, recent randomized and observational studies that coronary artery bypass grafting produces a modest, long-term survival advantage in stable patients undergoing revascularization, compared with percutaneous coronary intervention. The new finding seemed notable because it came from an unprecedentedly large U.S. dataset that broadly represented U.S. practice in the mid 2000s, and because the analysis used sophisticated statistical corrections and tests to try to eliminate patient-selection factors as confounders.

ASCERT, the ACCF and STS Database Collaborative on the Comparative Effectiveness of Revascularization Strategies, "is larger than any previous study [comparing coronary artery bypass surgery and percutaneous coronary intervention], it’s contemporary data [from cases treated during 2004-2007], and it’s more generalizable because it covers the entire United States," although limited exclusively to Medicare patients who were at least 65 years old, said Dr. William S. Weintraub, lead investigator of the study, at the annual meeting of the American College of Cardiology.

But several cardiologists who heard the report cautioned that despite all the analytic efforts by Dr. Weintraub and his associates to craft a comparison that eliminated biases, the results had limited implications because the analysis had an inherent inability to provide a truly clean comparison of coronary artery bypass surgery (CABG) and percutaneous coronary intervention (PCI) due to its reliance on observational data.

"It’s fair to say that at least in this data set, it’s impossible to eliminate residual confounders or selection bias because of the factors that a cardiologist or surgeon integrate at the bedside, such as frailty, patient preference, bleeding risk, and compliance with medical therapy," commented Dr. Alice K. Jacobs, a cardiologist and professor of medicine at Boston University. "I think these results will probably not change clinical practice."

"You can do multivariate, propensity analyses till kingdom come, but it will never eliminate the selection biases," in these data, said Dr. Bernard J. Gersh, a cardiologist and professor of medicine at the Mayo Clinic in Rochester, Minn.

"A major flaw in ASCERT was that it compared apples to oranges. It looked at a specific subset of very sick, older patients, and it did not take into account a variety of factors that could have affected these data," said Dr. Christopher J. White, president of the Society for Cardiovascular Angiography and Intervention in a written statement.

"PCI is not only a safe and effective treatment option for patients, it often is the only treatment available to patients too sick or too frail to have open heart surgery," added Dr. White, who is also professor and chairman for cardiovascular diseases at the Ochsner Heart and Vascular Institute in New Orleans.

Despite the limitations of the study, it may result in a small but meaningful resetting of how coronary revascularization is used, Dr. Weintraub said. "Practice has shifted toward PCI; this may change it back a little bit. But I don’t think it will have a tremendous effect by itself, and it probably shouldn’t," said Dr. Weintraub, chief of cardiology at Christina Care Health System in Newark, Del. "ASCERT largely supports the [coronary revascularization] guidelines that are already there," issued last year by the ACC, the American Association for Thoracic Surgery, and other organizations, he added (J. Am. Coll. Cardiol. 2011;58:2584-614; J. Am. Coll. Cardiol. 2011;58:2550-83).

"I hope we don’t have people saying that this shows that all patients should go to surgery," said cardiac surgeon Dr. Fred H. Edwards in an interview. "This is one more piece of information to use when evaluating how to treat patients with stable coronary artery disease. The information should be presented to patients and used in a dialog between surgeons and cardiologists," said Dr. Edwards, professor emeritus of surgery at the University of Florida in Jacksonville, director of the STS Research Center in Chicago, and coprincipal investigator for ASCERT.

"The key to ASCERT is its generalizability. In ASCERT we showed, hopefully definitively, that long-term survival in patients with stable two- and three-vessel coronary disease is better with surgery. But is survival the only thing that’s important? No, there is also the stroke rate, long-term quality of life, and patient preference."

The current ASCERT analysis focused on survival data for 86,244 patients who underwent CABG – data collected in the STS Adult Cardiac Surgery Database – and 103,549 patients who underwent PCI from the ACC Foundation’s National Cardiovascular Data Registry. The study’s central finding was that after adjustment by inverse probability weighing the 4-year mortality rate among CABG patients was 16.4%, and 20.8% among PCI patients. A second key finding was that initially after intervention, mortality was lower with PCI. The survival curves of the two treatment groups did not cross until 1 year out from the procedure, after which the survival advantage following CABG gradually increased over time.

Future reports will focus on other outcomes, including stroke rates, Dr. Weintraub said. Concurrent with his report at the meeting an article on the findings appeared online (N. Engl. J. Med. 2012 March 27 [doi:10.1056/NEJMoa1110717]).

Perhaps the most striking element in the findings was that the long-term survival benefit with CABG over PCI was consistent across a variety of subgroup analysis, including both women and men, and patients with or without diabetes.

"The advantage of CABG in all subgroups was a major surprise," Dr. Edwards said in an interview. "We thought that we’d see some subgroups that would benefit from PCI. Much to our surprise, all the subsets showed better survival with surgery, generally in the range of 20%-30%" on a relative basis.

Dr. Weintraub highlighted some extra analytic steps he and his associates took in the study. In addition to the primary adjusted analysis, the researchers performed a second, propensity-matched analysis on a subgroup of about 43,000 CABG patients and an equal number of PCI patients who closely matched for a list of over 20 demographic and clinical variables. The propensity-matched calculations found a similar long-term survival advantage for CABG.

They also examined whether the observed differences could be explained by an unmeasured confounder, such as frailty. To explain the observed between-group difference, "the unmeasured confounder would have to have an effect so large that it would have a hazard ratio of two, and would need to occur in about 30% of the patients in one group and in only 10% of those in the other group. How could we have missed that?" Dr. Edwards said.

Dr. Weintraub, Dr. Edwards, and Dr. White said they had no relevant disclosures. Dr. Jacobs said she had received research grants from Abiomed, Accumetrics, and Abbott Vascular. Dr. Gersh said he has been a consultant to Boston Scientific and Abbott Laboratories.

CHICAGO – Coronary artery bypass surgery had a significant 4-year survival advantage of 4 percentage points over percutaneous coronary intervention, based on analysis of an observational, U.S.-wide dataset with nearly 190,000 patients.

These results, from a new collaboration between the American College of Cardiology Foundation (ACCF) and the Society for Thoracic Surgeons (STS), confirmed findings made by other, recent randomized and observational studies that coronary artery bypass grafting produces a modest, long-term survival advantage in stable patients undergoing revascularization, compared with percutaneous coronary intervention. The new finding seemed notable because it came from an unprecedentedly large U.S. dataset that broadly represented U.S. practice in the mid 2000s, and because the analysis used sophisticated statistical corrections and tests to try to eliminate patient-selection factors as confounders.

ASCERT, the ACCF and STS Database Collaborative on the Comparative Effectiveness of Revascularization Strategies, "is larger than any previous study [comparing coronary artery bypass surgery and percutaneous coronary intervention], it’s contemporary data [from cases treated during 2004-2007], and it’s more generalizable because it covers the entire United States," although limited exclusively to Medicare patients who were at least 65 years old, said Dr. William S. Weintraub, lead investigator of the study, at the annual meeting of the American College of Cardiology.

But several cardiologists who heard the report cautioned that despite all the analytic efforts by Dr. Weintraub and his associates to craft a comparison that eliminated biases, the results had limited implications because the analysis had an inherent inability to provide a truly clean comparison of coronary artery bypass surgery (CABG) and percutaneous coronary intervention (PCI) due to its reliance on observational data.

"It’s fair to say that at least in this data set, it’s impossible to eliminate residual confounders or selection bias because of the factors that a cardiologist or surgeon integrate at the bedside, such as frailty, patient preference, bleeding risk, and compliance with medical therapy," commented Dr. Alice K. Jacobs, a cardiologist and professor of medicine at Boston University. "I think these results will probably not change clinical practice."

"You can do multivariate, propensity analyses till kingdom come, but it will never eliminate the selection biases," in these data, said Dr. Bernard J. Gersh, a cardiologist and professor of medicine at the Mayo Clinic in Rochester, Minn.

"A major flaw in ASCERT was that it compared apples to oranges. It looked at a specific subset of very sick, older patients, and it did not take into account a variety of factors that could have affected these data," said Dr. Christopher J. White, president of the Society for Cardiovascular Angiography and Intervention in a written statement.

"PCI is not only a safe and effective treatment option for patients, it often is the only treatment available to patients too sick or too frail to have open heart surgery," added Dr. White, who is also professor and chairman for cardiovascular diseases at the Ochsner Heart and Vascular Institute in New Orleans.

Despite the limitations of the study, it may result in a small but meaningful resetting of how coronary revascularization is used, Dr. Weintraub said. "Practice has shifted toward PCI; this may change it back a little bit. But I don’t think it will have a tremendous effect by itself, and it probably shouldn’t," said Dr. Weintraub, chief of cardiology at Christina Care Health System in Newark, Del. "ASCERT largely supports the [coronary revascularization] guidelines that are already there," issued last year by the ACC, the American Association for Thoracic Surgery, and other organizations, he added (J. Am. Coll. Cardiol. 2011;58:2584-614; J. Am. Coll. Cardiol. 2011;58:2550-83).

"I hope we don’t have people saying that this shows that all patients should go to surgery," said cardiac surgeon Dr. Fred H. Edwards in an interview. "This is one more piece of information to use when evaluating how to treat patients with stable coronary artery disease. The information should be presented to patients and used in a dialog between surgeons and cardiologists," said Dr. Edwards, professor emeritus of surgery at the University of Florida in Jacksonville, director of the STS Research Center in Chicago, and coprincipal investigator for ASCERT.

"The key to ASCERT is its generalizability. In ASCERT we showed, hopefully definitively, that long-term survival in patients with stable two- and three-vessel coronary disease is better with surgery. But is survival the only thing that’s important? No, there is also the stroke rate, long-term quality of life, and patient preference."

The current ASCERT analysis focused on survival data for 86,244 patients who underwent CABG – data collected in the STS Adult Cardiac Surgery Database – and 103,549 patients who underwent PCI from the ACC Foundation’s National Cardiovascular Data Registry. The study’s central finding was that after adjustment by inverse probability weighing the 4-year mortality rate among CABG patients was 16.4%, and 20.8% among PCI patients. A second key finding was that initially after intervention, mortality was lower with PCI. The survival curves of the two treatment groups did not cross until 1 year out from the procedure, after which the survival advantage following CABG gradually increased over time.

Future reports will focus on other outcomes, including stroke rates, Dr. Weintraub said. Concurrent with his report at the meeting an article on the findings appeared online (N. Engl. J. Med. 2012 March 27 [doi:10.1056/NEJMoa1110717]).

Perhaps the most striking element in the findings was that the long-term survival benefit with CABG over PCI was consistent across a variety of subgroup analysis, including both women and men, and patients with or without diabetes.

"The advantage of CABG in all subgroups was a major surprise," Dr. Edwards said in an interview. "We thought that we’d see some subgroups that would benefit from PCI. Much to our surprise, all the subsets showed better survival with surgery, generally in the range of 20%-30%" on a relative basis.

Dr. Weintraub highlighted some extra analytic steps he and his associates took in the study. In addition to the primary adjusted analysis, the researchers performed a second, propensity-matched analysis on a subgroup of about 43,000 CABG patients and an equal number of PCI patients who closely matched for a list of over 20 demographic and clinical variables. The propensity-matched calculations found a similar long-term survival advantage for CABG.

They also examined whether the observed differences could be explained by an unmeasured confounder, such as frailty. To explain the observed between-group difference, "the unmeasured confounder would have to have an effect so large that it would have a hazard ratio of two, and would need to occur in about 30% of the patients in one group and in only 10% of those in the other group. How could we have missed that?" Dr. Edwards said.

Dr. Weintraub, Dr. Edwards, and Dr. White said they had no relevant disclosures. Dr. Jacobs said she had received research grants from Abiomed, Accumetrics, and Abbott Vascular. Dr. Gersh said he has been a consultant to Boston Scientific and Abbott Laboratories.