Author + information
- Received September 8, 1998
- Revision received January 27, 1999
- Accepted March 26, 1999
- Published online July 1, 1999.
- Wayne B Batchelor, MD∗,
- Eric D Peterson, MD, MPH∗,* (, )
- Daniel B Mark, MD, MPH, FACC∗,
- J.David Knight, MS∗,
- Christopher B Granger, MD, FACC∗,
- Paul W Armstrong, MD, FACC† and
- Robert M Califf, MD, FACC∗
- ↵*Reprint requests and correspondence: Dr. Eric D. Peterson, Box 3236, Duke University Medical Center, Durham, North Carolina 27710
We sought to compare U.S. and Canada’s post-myocardial infarction (MI) cardiac catheterization practices in the detection of severe coronary artery disease (CAD).
Little is known about the efficiency with which the aggressive post-MI catheterization strategy observed in the U.S. detects severe CAD compared with the more conservative strategy observed in Canada.
From the U.S. and Canadian patients who had participated in the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries trial (n = 22,280, 11.5% Canadian), we examined the frequency of in-hospital cardiac catheterization, the prevalence of severe CAD observed at catheterization (diagnostic efficiency) and the total number of MI patients with severe CAD identified (diagnostic yield).
The rate of catheterization in the U.S. was more than 2.5 times that in Canada (71% vs. 27%, respectively, p < 0.001). With identical prevalences of severe CAD at catheterization (17%) in the two countries, the higher frequency of catheterization in the U.S. resulted in the identification of more than two and a half times as many cases of severe CAD compared with Canada (12 severe CAD cases identified per 100 post-MI patients in the U.S., vs. 4.6 per 100 in Canada). If considered in isolation, we estimated that these differences in severe disease detection might effect a small long-term survival advantage in favor of the U.S. strategy (estimated 5.0 lives saved per 1,000 MI patients).
Canada’s more restrictive post-MI cardiac catheterization strategy is no more efficient in identifying severe CAD than the aggressive U.S. strategy, and may fail to identify a substantial number of post-MI patients with high risk coronary anatomy. The long-term impact of these differences in practice patterns requires further evaluation.
Examination of the healthcare systems in the U.S. and Canada has become a standard for comparing patient outcomes between a technologically driven multipayer system (i.e., U.S.) and a more conservative, government-regulated healthcare system (i.e., Canada). Within the arena of cardiovascular healthcare, such comparisons have focused on differences in the management patterns after myocardial infarction (MI) (1–3). These studies have revealed that patients with MI in the U.S. are more likely to undergo diagnostic cardiac catheterization and revascularization procedures than are their Canadian counterparts. Despite reports suggesting gains in quality of life and functional status (1,3), more aggressive catheterization practices have not yet been shown to benefit short-term post-MI survival (1–8).
One of the primary purposes of catheterization after MI is to identify patients with severe coronary artery disease (CAD), such as triple-vessel disease (3VD) or left main coronary artery disease (LMD), who could achieve a survival benefit from surgical revascularization (9–12). With regard to the detection of severe CAD, different post-MI catheterization practice patterns can be evaluated by comparing their respective diagnostic efficiency (number of post-MI patients with severe CAD identified divided by the number of catheterizations performed), and diagnostic yield (total number of patients with severe CAD identified from the post-MI population). For example, by preferentially performing catheterization on patients with a high pre-test likelihood of severe CAD, a constrained healthcare system could maximize diagnostic efficiency by eliminating unnecessary catheterizations on those patients unlikely to have severe CAD. However, the same strategy, should it become too restrictive, might fail to identify some patients in the post-MI population with severe CAD, thereby reducing the overall diagnostic yield of the catheterization strategy.
The primary objective of this study was to compare U.S. and Canadian catheterization practices with respect to their diagnostic efficiency and yield in detecting severe CAD after MI. Second, to gain insight into the catheterization patterns within each of these healthcare systems, we examined the clinical characteristics predictive of cardiac catheterization and of severe CAD in the two countries. Our final objective was to estimate how any observed differences between U.S. and Canadian yields of severe CAD post-MI might affect long-term outcome.
We studied patients from the U.S. and Canada who were enrolled in the GUSTO-1 trial (13). The eligibility criteria and results of this trial, which compared four thrombolytic strategies for acute MI, have been reported previously (13). We considered the 26,003 patients randomized into the study from U.S. (n = 23,105) and Canadian (n = 2,898) sites. From this cohort, a total of 3,723 patients who either had a prior revascularization procedure (n = 2,517), or had participated in the GUSTO Angiographic Substudy (14)(a subgroup of randomly selected GUSTO patients all of whom underwent cardiac catheterization by protocol, n = 1,154), were excluded. However, patients from the angiographic substudy were studied separately to develop a model predictive of severe CAD based on baseline clinical characteristics (see below). The remaining 22,280 patients (19,739 U.S. and 2,541 Canadian patients), in whom cardiac catheterization was left to the discretion of the attending physician, formed our final study cohort.
Cardiac catheterization results and practice patterns
Baseline clinical characteristics and in-hospital cardiac catheterization results were derived from the GUSTO Trial Case Report Form (13). Severe CAD was defined as either the presence of a ≥70% stenosis in all three major epicardial coronary arteries, or a ≥50% stenosis in the left main coronary artery on visual angiographic assessment, as interpreted by the site physician performing the cardiac catheterization. By examining the rates and angiographic results of in-hospital post-MI cardiac catheterization, we calculated the diagnostic efficiency and diagnostic yield for the detection of severe CAD in the U.S. and Canada. To gain insights into clinical factors influencing the selection for cardiac catheterization, we examined each country’s rate of catheterization after stratification by age (both by decade, and after dichotomizing age into <75 vs. ≥75 years), gender, diabetes mellitus, MI location (anterior vs. other), prior MI, Killip class (I vs. ≥II), recurrent myocardial ischemia and occurrence of any post-MI complication (i.e., shock, pulmonary edema, congestive heart failure, recurrent ischemia, sustained ventricular tachycardia, ventricular fibrillation or recurrent myocardial infarction). These rates of catheterization were then compared to the group’s corresponding likelihood of severe CAD to relate cardiac catheterization practice patterns to the risk of severe CAD.
Estimating the prevalence of severe CAD
Comparisons between U.S. and Canadian cardiac catheterization practices should ideally correct for the underlying prevalence of severe CAD in the two countries. We estimated each country’s prevalence of severe CAD in two ways. First, we directly observed the prevalence of severe CAD in the U.S. and Canadian patients (n = 1,154, 14% Canadian) from the GUSTO Angiographic Substudy (14). Second, using the clinical and angiographic data from this substudy, we developed a logistic regression model predicting the likelihood of severe CAD based on baseline clinical characteristics. Clinical variables demonstrating a significant univariate association with the presence of severe CAD (p < 0.05) were entered, in a forward stepwise fashion, into a multivariable model with the presence of severe CAD modeled as the dependent variable. This model was then applied to our cohorts in the U.S. and Canada to determine whether the overall predicted probability of severe CAD differed between patients from these two countries.
Potential impact on long-term outcome
If one healthcare system identifies and revascularizes more post-MI patients with severe CAD than another, then we might anticipate the former system to effect a long-term survival benefit relative to the other healthcare system (12). To estimate this potential survival difference, we assumed that patients identified as having severe CAD in the two countries would be equally likely to undergo surgical revascularization (at a rate consistent with that observed in the GUSTO trial). We then applied the marginal benefit of surgical revascularization over medical therapy reported by Yusuf et al. (12)to estimate the five-year survival gradient that might accrue from any measured difference in severe CAD yield between the two countries’ catheterization practices. It must be recognized that the sole purpose of these calculations was to achieve a gross estimate of the potential long-term effects of the differences in these two different catheterization strategies when considered in isolation of other management practices.
Baseline clinical characteristics
Overall, the demographic and clinical characteristics of the U.S. and Canadian patients were similar (Table 1). The U.S. cohort had a slightly higher proportion of women, hypertension, diabetes, hypercholesterolemia and patients with a family history of coronary artery disease, and a slightly higher heart rate on presentation. Canadian patients had a higher baseline mean systolic blood pressure and were slightly more likely to have had prior angina or MI.
Cardiac catheterization findings
In-hospital post-MI cardiac catheterization was performed more than two and a half times more frequently in the U.S. than in Canada (71% vs. 27%, respectively, p < 0.001). Despite this, the findings at catheterization were nearly identical; the proportion of patients with no significant coronary stenoses and those with one-, two- and three-vessel disease, and left main disease were similar in the two countries (Table 2, p = NS for all comparisons). Accordingly, the diagnostic efficiency with which severe CAD was detected was also equal for the two countries (17% of catheterized patients having severe CAD in both the U.S. and Canada). However, due to the higher rate of catheterization in the U.S., the diagnostic yield for severe CAD in the U.S. was more than two and a half times that in Canada (12 cases of severe CAD identified per 100 MI patients in the U.S. compared with 4.6 per 100 in Canada, p < 0.001). Alternatively stated, for every 100 MI patients treated, the U.S. post-MI management strategy identified an average of 7.4 more cases of severe CAD than did Canadian practices (Fig. 1), at the expense of performing 44 more catheterizations.
To ensure that these intercountry differences in the identification of severe CAD did not result from a difference in underlying disease prevalence, we estimated each country’s prevalence of severe CAD using two methods. First, we examined the prevalence of severe CAD in the two countries as measured directly from the GUSTO Angiographic Substudy (14)(Table 3). In this angiographic substudy, the frequency of severe CAD was similar for U.S. and Canadian patients (19% vs. 22%, respectively, p = NS). Second, we estimated each country’s underlying likelihood for severe disease based on their respective baseline risk factors. Multivariable logistic regression identified age, previous MI, male gender, previous coronary bypass surgery, previous percutaneous coronary angioplasty, hyperlipidemia and smoking history as independent clinical predictors of severe CAD (see Appendix). The C-index for the predictive model was 0.72. Applying this model to our entire sample produced a predicted prevalence of severe CAD of 18.7% for U.S. patients, and 18.8% for Canadian patients (p = NS). Thus, with both direct observation and indirect estimates, we demonstrated that the underlying prevalence of severe CAD was similar among the patients enrolled in the two countries.
Relationship between cardiac catheterization rates and the likelihood of severe CAD in clinical subgroups
Comparisons of U.S. and Canadian cardiac catheterization rates for each of the defined clinical subgroups are shown in Table 4. United States physicians were two to three times more likely than Canadian physicians to use cardiac catheterization across a wide spectrum of patients. For each clinical subgroup, we compared the likelihood of underlying severe CAD juxtaposed against the likelihood of undergoing catheterization in the U.S. and Canada (Fig. 2). Of interest, several predictors of severe CAD (i.e., increased age, diabetes, prior MI and higher Killip class) were paradoxically associated with a lower likelihood of cardiac catheterization in both countries. To further illustrate this paradox, we specifically displayed the relationship between the predicted probability of severe CAD and rates of cardiac catheterization across various age decades (Fig. 3). Although catheterization was performed with significantly higher frequency in the U.S. across all age strata, the likelihood for cardiac catheterization in both countries fell dramatically with increasing age. This pattern was in direct contrast to these patients’ corresponding likelihood for severe CAD. For example, although over 80% of U.S. patients of age ≤50 years underwent catheterization, the probability of severe CAD in these young patients was only approximately 10%. In contrast, only one half of elderly patients (i.e., age 76 to 85 years) in the U.S., and less than one quarter in Canada, underwent catheterization, yet these patients had a nearly 30% probability of severe CAD. Therefore, in Canada, the rate of cardiac catheterization in the elderly was actually less than their corresponding probability of severe CAD.
Given the observed differences in diagnostic yield, the more aggressive U.S. catheterization strategy identifies 48 more patients with 3VD and 26 more with LMD per 1,000 MI patients than would be identified in Canada (Table 5). Assuming that at least 50% of these patients are eligible for, and undergo, surgical revascularization (consistent with the observed coronary artery bypass graft rate in the GUSTO trial), approximately 24 and 13 more patients with 3VD and LMD, respectively, undergo revascularization in the U.S. relative to Canada. Yusuf has reported that an initial strategy of coronary artery bypass surgery affords a five-year mortality odds ratio of 0.32 (95% confidence interval [CI] of 0.15 to 0.70) for LMD, and 0.58 (95% CI of 0.42 to 0.80) for 3VD compared with those initially treated with medical therapy (12). Therefore, we estimate that the five-year survival advantage that might accrue from these marginal differences in severe CAD yield amounts to 5.0 lives saved per 1,000 MI patients, in favor of the U.S. strategy (Table 5).
This observational study is the first to compare post-MI cardiac catheterization practices in the U.S. and Canada regarding their respective abilities to detect severe CAD after MI. Our findings suggest that the in-hospital post-MI catheterization strategies in both countries are not ideally designed to target patients with a higher likelihood of severe CAD who may benefit from myocardial revascularization. As a result, the diagnostic efficiency with which severe CAD is detected in both countries is similar, and is in fact no better than one might anticipate from a merely random selection process. With no better efficiency than the U.S. practice, (and a much lower rate of catheterization), the Canadian strategy identifies substantially fewer patients with severe CAD in the post-MI hospital phase. We estimate that the long-term effect of these differences might be significant, albeit modest (an estimated 5.0 lives saved per 1,000 MI patients in favor of the U.S. strategy) when considered in isolation from other differences in post-MI management.
Previous studies comparing invasive versus noninvasive post-MI strategies
Prior studies demonstrating major variances in both regional and international cardiac catheterization practices have generated a great deal of interest in studying whether such variances might affect post-MI outcomes (1–8,15). These studies have been inconclusive in determining whether there is a definitive benefit with a more aggressive catheterization strategy. In two previous comparisons of U.S. and Canadian practices, Rouleau et al. (1)and Mark et al. (3)have reported benefits in angina symptoms and quality of life associated with the more aggressive cardiac catheterization strategy of the U.S. However, other studies including several observational studies (2,7,8), a systematic overview (4), the Veterans Affairs Non–Q-wave Infarction Strategies In-Hospital trial (VANQWISH) (6)and the Organization for the Assessment of Strategies for Ischemic Syndromes Study (OASIS) (16)have suggested that there is no significant benefit to a more invasive management strategy for MI and unstable coronary syndromes. However, because the eventual survival advantage for coronary artery bypass surgery over medical therapy may become evident only after several years, any study comparing different catheterization strategies requires several years of follow-up (12,17). Only two reports (7,8)have had both adequate power and long-term follow-up to examine these effects on survival. Although neither of these studies showed a difference in survival between conservative and aggressive catheterization strategies, the differences in catheterization frequency and revascularization between the aggressive and conservative strategies in these reports were substantially less than observed in our GUSTO U.S.–Canadian comparison.
The cardiac catheterization paradox: insights from studying catheterization practice patterns
To gain insights into the potential long-term implications of these differences in cardiac catheterization strategies, our study compared the efficiency and overall effectiveness (diagnostic yield) with which Canadian and U.S. post-MI testing patterns identify severe CAD. Such an approach is based on the premise that the mortality benefit of surgical revascularization is highly dependent on the identification of such high risk patients by cardiac catheterization and coronary angiography. We anticipated that, due to more restricted use, cardiac catheterization in Canada would be reserved for MI patients deemed from clinical or noninvasive risk stratification to be at higher risk for severe CAD, thereby increasing Canada’s diagnostic efficiency. However, this was not the case. The two countries shared remarkably similar patterns of patient selection for cardiac catheterization—patterns that were not designed to target patients with a higher likelihood of severe CAD. In fact, paradoxically, there was a tendency in both countries to perform less catheterization on MI patients who are known to be at the highest risk (18)and most likely to have underlying severe CAD (elderly patients, diabetics and patients with previous MI and congestive heart failure). Conversely, catheterization rates were much higher in younger, lower risk patients who were unlikely to have severe CAD. As a result of these paradoxical trends in the application of cardiac catheterization, the U.S. and Canada shared similar inefficiency in severe CAD detection. In fact, given an estimated severe CAD prevalence of approximately 19% in both countries, a diagnostic efficiency of 17% in both countries suggests that the selection process for catheterization in the U.S. and Canada is no better in targeting severe CAD than a completely random selection. These paradoxic practice patterns attest to the complexities in the decision to perform catheterization. Such a decision has been shown to be strongly influenced by a number of nonclinical factors including geographic region, local cardiac catheterization lab availability, financial influences, fear of litigation and patient expectations (5,19–21).
Implications of the cardiac catheterization rate–efficiency relationship
Under circumstances of equivalent efficiency, the higher rate of catheterization in the U.S. results in a more than twofold higher proportion of patients with severe CAD being identified in the U.S. compared with Canada (∼12 cases of severe CAD per 100 MI patients identified in the U.S. compared with ∼5 per 100 MI patients in Canada). Interestingly, Tu et al. (22), in comparing rates of coronary bypass surgery, have also alluded to a more than twofold greater detection of severe disease in the U.S. compared with Canada. Therefore, healthcare systems with lower rates of cardiac catheterization such as Canada’s would have to increase the efficiency of catheterization (i.e., optimize catheterization case selection) to maintain similar diagnostic yields of severe CAD to higher rate healthcare systems such as the U.S. It remains to be determined whether the use of clinical predictors for severe CAD (23), or incremental predictive information from noninvasive testing might improve patient selection, thereby achieving this goal for lower rate systems.
There are several limitations to our study. First, when evaluating post-MI prognosis, cardiac catheterization comparisons cannot be interpreted in isolation. Important prognostic information, independent of coronary anatomy, can also be derived from noninvasive post-MI cardiac testing, the results of which were not available from our study (24,25). Second, it is impossible to definitively ascertain the underlying prevalence of severe CAD in the post-MI population from either country. However, using both direct observations from the GUSTO Angiographic Substudy and logistic regression modeling, we do predict similar prevalences. Third, because angiograms were interpreted visually by the physician performing the procedure, there exists a potential for interobserver variability in assessing coronary artery disease severity. However, all grades of stenosis severity, from no significant disease to severe CAD categories, were similar in the two countries, and reanalyzing our results using a criterion of ≥50%, instead of ≥70% stenosis as the definition of significant 3VD did not significantly alter the U.S.–Canadian differences in severe CAD yield (other than increasing the yield equally in the two countries). Fourth, we examined only in-hospital catheterization rates. Because there may be a substantial delay in acquiring cardiac catheterization in Canada, it is possible that more of these procedures were done on an outpatient basis. However, several trials suggest that outpatient post-MI cardiac catheterization rates differ minimally between the two countries (2,3,26). In the Economic and Quality of Life Substudy of the GUSTO trial, outpatient rates of catheterization at 1 year post-MI were similar between the U.S. and Canada (23% vs. 19%, respectively, p = NS). Therefore, it is unlikely that outpatient rates of catheterization in Canada would have made up for the large (2.5-fold) inpatient differences in severe CAD yield. Fifth, it must be recognized that our results reflect only the average catheterization strategies in the two countries; they are not sensitive to the major regional differences in catheterization practices that exist within the U.S. and Canada (3,5). Finally, our projected influence on long-term survival used a weighted estimate of survival differences between coronary artery bypass surgery and medical therapy based on clinical trials data between 1972 and 1984 (12). Advances in medical therapy and surgical technique, as well as the increased use of percutaneous coronary interventions (angioplasty) render an estimate of the current advantage of revascularization (bypass surgery or angioplasty) over medical therapy for severe CAD more complex. Given these limitations, the true impact of identifying severalfold more patients with severe CAD with U.S. catheterization practices remains speculative.
Our study provides new insights into the diagnostic properties of two different cardiac catheterization strategies. More restricted catheterization practices, such as those in Canada, if not accompanied by increased diagnostic efficiency, identify substantially fewer patients with high risk coronary anatomy than the more aggressive U.S. practice. These differences could have modest, but significant implications on long-term post-MI outcomes. Both countries might substantially improve the efficiency of post-MI management by more effectively targeting higher risk patient subgroups for cardiac catheterization. A better understanding of the effects of these differences in catheterization utilization, particularly within the context of independent prognostic information derived from noninvasive post-MI risk stratification, will require long-term study, and is crucial to optimizing post-MI healthcare delivery in an era of constrained financial resources.
We thank Tracey Simons for her editorial assistance with preparation of the manuscript.
Risk model for severe CAD
Probability of severe CAD = 1/[1 + exp (−L)], where L = −3.822 + 0.0376 (Age) − 0.6058 (Gender) − 0.3645 (HxSmk) + 0.3553 (Lipids) + 1.0252 (Prev MI) + 1.7153 (Prev CABG) − 0.8527 (Prev PTCA).
1. For Gender: male = 0, female = 1.
2. HxSmk: history of cigarette smoking (smoker = 1, nonsmoker = 0).
3. Lipids: history of hypercholesterolemia (yes = 1, no = 0).
4. Prev MI: history of previous myocardial infarction (yes = 1, no = 0).
5. Prev CABG: previous coronary artery bypass surgery (yes = 1, no = 0).
6. Prev PTCA: previous percutaneous transluminal coronary angioplasty (yes = 1, no = 0).
☆ Dr. Batchelor is supported by a Junior Research Fellowship from the Heart and Stroke Foundation of Canada.
- coronary artery disease
- confidence interval
- Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Arteries
- left main coronary disease
- myocardial infarction
- three-vessel coronary disease
- Received September 8, 1998.
- Revision received January 27, 1999.
- Accepted March 26, 1999.
- American College of Cardiology
- Michels K.B,
- Yusuf S
- Boden W.E,
- O’Rourke R.A,
- Crawford M.H,
- et al.
- Every N.R,
- Parsons L.S,
- Fihn S.D,
- et al.
- CASS Principal Investigators and their associates
- Ferguson J.J
- GUSTO Investigators,
- Tardiff B.E,
- Califf R.M,
- Morris D,
- et al.
- GUSTO-1 Investigators,
- Lee K.L,
- Woodlief L.H,
- Topol E.J,
- et al.
- Task Force on Practice Guidelines,
- Ryan T.J