COURAGE Results Consistent Across Different Health Care Systems
Key Points:
Outcomes largely similar for PCI vs. medical therapy in different health care systems
Possible mortality reduction with PCI in highest risk patients
By Caitlin E. Cox
Friday, July 30, 2010
Although the COURAGE trial enrolled patients from 3 distinct North American health care systems, each with different levels of baseline risk, outcomes largely were similar for optimal medical therapy vs. percutaneous coronary intervention (PCI) within the patient subgroups. Any discrepancies can be explained by baseline differences known to affect long-term prognosis, asserts a paper published online July 27, 2010, in Circulation: Cardiovascular Quality and Outcomes.
Yet Jeffrey W. Moses, MD, of Columbia University Medical Center (New York, NY), described the paper as “defensive.” In a telephone interview with TCTMD, he claimed the analysis demonstrates that the highest risk patients actually appear to have a mortality benefit with the addition of PCI to optimal medical therapy.
Many Faces of COURAGE
The COURAGE (Clinical Outcomes Utilizing Revascularization and Aggressive druG Evaluation) trial, which randomized 2,287 stable CAD patients to optimal medical therapy with or without PCI, found no significant difference between the 2 approaches for the primary endpoint of all-cause death and nonfatal MI or for secondary composite endpoints.
For the current study, Bernard R. Chaitman, MD, of Saint Louis University School of Medicine (St. Louis, MO), analyzed the results of COURAGE patients based on their place of enrollment:
15 US Department of Veterans Affairs (VA) sites (n = 968)
19 US non-VA sites (n = 386)
16 Canadian sites (n = 931)
Within each health care system, baseline characteristics were similar between PCI and optimal medical therapy groups. Yet among the 3 types of health care systems, there were different levels of risk.
The VA group generally was at highest risk, with more patients taking beta-blockers, calcium antagonists, nitrates, and diuretics at baseline and throughout the study compared with the other 2 groups. They also were less likely to have met target HDL levels (> 40 mg/dL). The Canadian group, meanwhile, was more likely to have entered the study having already achieved treatment targets for hypertension, diet, and physical activity. The non-VA group tended to be at intermediate risk between the Canadian and VA patients.
Compared with US patients in general, Canadian patients were younger, had lower BMIs, were more likely to be white, and more regularly performed moderate exercise. They also were more likely to have experienced prior MI or revascularization but less likely to have a history of hypertension or stroke.
At a median of 4.6-year follow-up, both all-cause death/MI and death alone were similar between optimal medical therapy and PCI among the 3 health care systems (table 1).
Table 1. Outcomes Based on Place of Enrollment
Optimal Medical Therapy
Optimal Medical Therapy + PCI
P Value
All-Cause Death/MI
VA
Non-VA
Canadian
21.9%
21.8%
13.5%
22.3%
15.8%
17.3%
0.95
0.24
0.17
Death
VA
Non-VA
Canadian
11.4%
8.3%
5.0%
8.6%
7.5%
6.9%
0.07
0.94
0.37
Logistic regression analysis initially showed a significant interaction between type of health care system and death, with the VA group having higher mortality compared with the Canadian group (OR 1.92; 95% CI 1.34-2.75; P < 0.001). However, adjustment for baseline characteristics somewhat blunted that relationship (adjusted OR 1.55; 95% CI 1.02-2.35; P = 0.04). Other independent predictors of mortality were baseline ejection fraction, smoking status, age, and number of comorbidities.
Analysis Offers ‘Unique Opportunity’
“Our main findings are that type of [health care system] did not influence the primary outcome,” or any other endpoint, when optimal medical therapy and PCI were compared, Dr. Chaitman and colleagues conclude. “A significant difference in mortality was observed at 4.6 years of follow-up among the different [health care systems] that was largely explained by age, comorbid conditions, and ventricular function.”
The investigators say their analysis “provides a unique opportunity to explore treatment differences over a gradient of comorbidity and different processes of care in a large cohort of patients with stable ischemic heart disease treated in a contemporary fashion. The results amplify the overall COURAGE findings and reveal that there are no significant treatment differences between PCI and optimal medical therapy for serious cardiovascular outcomes across different [health care systems].”
They acknowledge several study limitations, among them the lack of power to test for interactions because each health care system represents a subset of the overall trial. In addition, because the optimal medical therapy offered in COURAGE involved free medication and an aggressive approach, the potential to observe differences “was likely minimized compared to what might have been observed with a usual care approach,” the authors write, noting that the trial provides a model for implementing optimal medical therapy in diverse environments.
Missing the Point
Dr. Moses, however, had a different take-away message from the results. “This is a defensive paper,” he commented.
“The most striking thing here is that they make a pretty compelling case that the VA group was the sickest, the US non-VA group was the next sickest, and the Canadian group was the least sick,” he said, pointing out that in the VA group the mortality difference nearly achieved statistical significance at a P value of 0.07.
The COURAGE investigators “hang on to the non-statistical significance, which is naive. Whenever you’re doing these analyses, they’re exploratory by definition,” Dr. Moses said. “When you’re dealing with an underpowered trial [as is the case with COURAGE] and then you do subgroup analyses and see a P value of 0.07, it should at least raise a flag. To see a trend this strong is interesting.”
“Importantly, and this is the more profound point, every analysis of COURAGE where they stratified people on the basis of risk—ie, when they looked at ischemia, multivessel disease with low ejection fraction, silent ischemia—all of them showed trends toward reductions in hard events,” with PCI in moderately high risk patients, Dr. Moses said. He stressed that the study did not include extremely high-risk patients.
Moreover, the mortality difference is not trivial, Dr. Moses stressed. “It’s a little frustrating because there’s a lot in COURAGE that, while not definitive, explains why PCI is important in terms of events and conforms with 30 years of our understanding of risk stratification,” he said. “They just keep burying that message. I guess they think that if they just keep saying it, it will be true.”