The flow diagram details the creation of the study cohort and the corresponding rates of
obstructive CAD rates in New York and Ontario. Patients may have met more than 1 exclusion
criterion. CABG indicates coronary artery bypass graft; MI, myocardial infarction.
eTable 1. Relative importance of explanatory factors in predicting the presence of obstructive
coronary artery disease in New York State and Ontario
eFigure 1. Observed vs predicted probability of obstructive coronary artery disease in New York
State and Ontario
Ko DT, Tu JV, Austin PC, Wijeysundera HC, Samadashvili Z, Guo H, Cantor WJ, Hannan EL. Prevalence and Extent of Obstructive Coronary Artery Disease Among Patients
Undergoing Elective Coronary Catheterization in New York State and Ontario. JAMA. 2013;310(2):163-169. doi:10.1001/jama.2013.7834
Prior studies have shown that physicians in New York State (New York) perform twice as many
cardiac catheterizations per capita as those in Ontario for stable patients. However, the role of
patient selection in these findings and their implications for detection of obstructive coronary
artery disease (CAD) are largely unknown.
To evaluate the extent of obstructive CAD and to compare the probability of detecting obstructive
CAD for patients undergoing cardiac catheterization.
Design, Setting, and Patients
An observational study was conducted involving patients without a history of cardiac disease who
underwent elective cardiac catheterization between October 1, 2008, and September 30, 2011.
Obstructive CAD was defined as diameter stenosis of 50% or more in the left main
coronary artery or stenosis of 70% or more in a major epicardial vessel.
Main Outcomes and Measures
Observed rates and predicted probabilities of obstructive CAD. Predicted probabilities were
estimated using logistic regression models.
A total of 18 114 patients from New York and 54 933 from Ontario were included. The
observed rate of obstructive CAD was significantly lower in New York at 30.4% (95% CI, 29.7%-31.0%)
than in Ontario at 44.8% (95% CI, 44.4%-45.3%; P < .001). The
percentage of patients with left main or 3-vessel CAD was also significantly lower in New York than
in Ontario (7.0% [95% CI, 6.6%-7.3%] vs 13.0% [95% CI, 12.8%-13.3%];
P < .001). In New York, a substantially higher percentage of
patients with low predicted probability of obstructive CAD underwent cardiac catheterization; for
example, only 19.3% (95% CI, 18.7%-19.9%) of patients undergoing cardiac catheterization in New York
had a greater than 50% predicted probability of having obstructive CAD than those in Ontario at 41%
(95% CI, 40.6%-41.4%; P < .001). At 30 days, crude mortality for
patients undergoing cardiac catheterization was slightly higher in New York at 0.65% (90 of
13 824; 95% CI, 0.51%-0.78%) than in Ontario at 0.38% (153 of 40 794; 95% CI,
0.32%-0.43%; P < .001).
Conclusions and Relevance
In Ontario compared with New York State, patients undergoing elective cardiac catheterization
were significantly more likely to have obstructive CAD. This appears to be related to a higher
percentage of patients in New York with low predicted probability of CAD undergoing cardiac
The continuing increase in health care expenditures is threatening the sustainability of the
health care system and the economy of many developed countries.1- 3 Debates among the public, physicians, funders, and
policy makers have concentrated on how to provide better quality of care at a lower cost.4 In the United States, a study found that only 1 in 3 patients
who received elective cardiac catheterization had obstructive coronary artery disease (CAD), which
raises concerns about the necessity of cardiac procedures for many patients with stable CAD.5 According to these findings, one might reasonably conclude that
a more selective use of cardiac catheterization should be implemented to reduce its associated cost
and to improve its diagnostic efficiency. Conversely, other studies have suggested that a
restrictive approach is not more efficient in identifying severe CAD and could actually lead to an
underdiagnosis of patients who may benefit from coronary revascularization.6,7
Previous cross-country comparison studies between the United States and Canada have highlighted
large differences in the utilization of cardiac procedures because of different methods of
incentivizing health care.8- 10 Our
group has previously shown that clinicians in New York State (New York) perform twice as many
cardiac catheterizations per capita as are performed in Ontario, which could be explained by a
difference in the burden of CAD or by a difference in the patient selection process for
procedures.11 Given the increasing focus on how best to use
scarce health care resources, it is important to understand the reasons underlying the different
utilization patterns and their associated implications. Accordingly, the first objective of the
present study was to evaluate the extent of obstructive CAD in these jurisdictions. The second
objective was to evaluate whether there were different thresholds for selecting patients for cardiac
catheterization by comparing the predicted probabilities of obstructive CAD in New York and
The New York Cardiac Catheterization Database was used to evaluate patients undergoing cardiac
catheterization.12 As previously described, it is a
voluntary data system maintained by the state’s Department of Health. The database collects
information on demographics, medical comorbidities, cardiac conditions, ischemic testing, and
coronary anatomy among 18 participating hospitals with cardiac catheterization facilities. The
database was then linked to the Percutaneous Coronary Intervention (PCI) Reporting System and the
Cardiac Surgery Reporting System to determine rates of coronary revascularizations after index
cardiac catheterization.13 The Social Security
Administration Death Master File was used to identify deaths.13
This study was approved by the Sunnybrook Health Sciences Centre research ethics board.
The Cardiac Care Network of Ontario maintains an ongoing prospective clinical registry of all
patients undergoing cardiac catheterizations, as well as other invasive cardiac procedures in the
province. It has been demonstrated to be a valuable source of information in clinically oriented
research and has been used extensively.14 Both the New York
and Ontario databases are clinical data sets that contain similar data elements with standardized
data definitions that are suitable for comparison. We used the Canadian Institutes for Health
Information Database to capture coronary revascularization rates and the Registered Persons Database
to capture deaths after cardiac catheterization in Ontario.14
All adult patients in Ontario older than 20 years without a history of cardiac disease and who
underwent cardiac catheterization between October 1, 2008, and September 30, 2011, for stable CAD
were eligible for inclusion. In New York, the study sample was drawn from 18 of 82 cardiac
catheterization hospitals that participated in the Cardiac Catheterization Database. We identified
patients without a history of cardiac disease undergoing elective cardiac catheterization in a
sequential manner as proposed by Patel and colleagues5
(Figure). For patients who had multiple cardiac
catheterizations during the study period only the first cardiac catheterization was considered.
We defined obstructive CAD as stenosis of 50% or more of the left main coronary
artery or stenosis of 70% or more of a major epicardial or branch vessel. The presence of 3-vessel
CAD was defined by stenosis of 70% or more in the left anterior descending coronary artery, left
circumflex coronary artery, and right coronary artery.
We compared demographic characteristics, clinical characteristics, the extent of obstructive CAD,
and revascularization and mortality rates among patients who underwent elective cardiac
catheterization using χ2 tests for categorical variables and t
tests for continuous variables. Due to privacy restrictions that limited the transfer of data out of
each jurisdiction, statistical calculations were performed without merging the data sets.
To understand potential differences in how patients were selected for cardiac catheterization
between New York and Ontario, we first constructed logistic regression models in Ontario to predict
the presence of obstructive CAD. Selection of predictor variables was based on clinical knowledge
and prior literature.5,15 Variables included in
our model were age, sex, cardiac risk factors (diabetes, hyperlipidemia, smoking status,
hypertension), comorbidities (peripheral vascular disease, cerebrovascular disease, heart failure,
dialysis), Canadian Cardiovascular Society (CCS) angina classification, and high-risk ischemia
evaluation on noninvasive imaging.
Discrimination ability of the models was determined using the area under the receiver operating
characteristics curve (C statistic). No variable in the prediction model had an associated variance
inflation factor greater than 5, suggesting no multicollinearity. After model estimation, predicted
rates of obstructive CAD for each patient in New York and Ontario were calculated using the
coefficient estimates obtained in the Ontario cohort. This method is analogous to direct
standardization, which allowed us to estimate the expected probability of a given patient having
obstructive CAD if that patient had resided and managed in Ontario.8,9,16 The predicted probability of obstructive CAD of each
patient in each jurisdiction was then aggregated at the hospital level or the regional level to
allow comparison between New York and Ontario. Therefore, an individual patient was the unit of
analysis, and results were then aggregated to calculate average predicted probability of obstructive
CAD. We examined the calibration of the predicted probabilities using graphical calibration plots as
previously described.17,18 To do so, we
graphically compared predicted vs observed probabilities of the presence of obstructive CAD across
the deciles of risk.
SAS version 9.2 (SAS Institute Inc) was used for statistical analyses. A 2-sided
P value of .05 or less was considered statistically significant in the comparison
The creation of the study cohort is shown in the Figure. We included 61 756 patients in New York and 160 563 patients in Ontario
who underwent cardiac catheterization between October 1, 2008, and September 30, 2011. We excluded
3854 patients in New York and 8636 in Ontario with prior valvular disease, 6462 patients in New York
and 13 208 in Ontario with prior myocardial infarction, 18 117 patients in New York and
32 285 in Ontario with previous coronary revascularizations (PCI and CABG surgery), and 6102
patients in New York and 35 067 in Ontario with urgent or emergent indications (cardiogenic
shock and myocardial infarction) for cardiac catheterization. Our final cohort included 18 114
patients in New York and 54 933 patients in Ontario, representing 29.3% and 34.2% of the
The Figure also details rates of obstructive CAD. Prior to applying any exclusion criteria,
obstructive CAD was detected by cardiac catheterization in 50.6% (95% CI, 50.2%-51.0%) of patients
in New York and 61.8% (95% CI, 61.6%-62.1%) of patients in Ontario
(P < .001). After applying sequential exclusions to identify
patients with elective procedures without prior heart disease, the rate of obstructive CAD was still
lower in New York (30.4%; 95% CI, 29.7%-31.0%) than in Ontario (44.8%; 95% CI, 44.4%-45.3%;
P < .001).
Differences in demographics and clinical characteristics were observed among patients undergoing
cardiac catheterization (Table 1). Patients in New
York were significantly younger (mean, 61.2 [SD, 12.4] years vs 63.7 [11.4] years) and more
likely to be women (45.3% vs 39.0%) than those in Ontario. A higher proportion of asymptomatic
patients without typical angina as categorized by the Canadian Cardiovascular Society class 0 was
noted in New York (57.7% vs 29.3% in Ontario). Noninvasive ischemic testing prior to cardiac
catheterization was performed more often in Ontario than in New York (75.1% vs 63.2%;
P < .001). Among patients who underwent noninvasive testing, the
proportion of patients with high-risk findings on ischemic evaluation was substantially lower in New
York (4.7%) than in Ontario (50.9%; P < .001). Hospital
characteristics of patients who underwent cardiac catheterization differed significantly between New
York and Ontario with 56.9% vs 73.2% of patients who received cardiac catheterization at
full-service hospitals with capability to perform PCI and coronary artery bypass graft surgery
(CABG) (Table 1).
The anatomic results of the cardiac catheterizations are shown in Table 2. In New York, 2.5% (95% CI, 2.3%-2.8%) of patients who underwent
cardiac catheterization were found to have left main stenosis, 5.2% (95% CI, 4.9%-5.5%) had 3-vessel
CAD, and 7.0% (95% CI, 6.6%-7.3%) had left main or 3-vessel disease. In Ontario, patients were
significantly more likely to have severe CAD; 5.0% (95% CI, 4.9%-5.2%) had left main stenosis, 9.8%
(95% CI, 9.6%-10.1%) had 3-vessel coronary artery stenosis, and 13.0% (95% CI, 12.8%-13.3%) had left
main or 3-vessel disease (all P < .001).
Factors predicting the presence of obstructive CAD were found to have similar coefficient
estimates, suggesting that they are of similar importance in both jurisdictions (eTable 1 in Supplement). The Ontario regression
model had a C statistic of 0.74 to predict the presence of obstructive CAD. We then assessed its
performance using New York data as a validation or test sample. When the coefficient estimates
obtained from the Ontario data were applied to the New York data, the C statistic was 0.70,
suggesting generalizability of the model. We also examined the calibration of the predicted
probabilities by comparing predicted vs observed probabilities of the presence of obstructive CAD
across the deciles of risk, which demonstrated strong concordance between observed and predicted
rates in the calibration plot (eFigure 1 in Supplement).
Table 3 depicts the proportion of patients
undergoing cardiac catheterization stratified by predicted probability of obstructive coronary
disease based on the model fitted to the Ontario cohort. The results presented in Table 3 demonstrated that patients who received cardiac
catheterization in New York had a significantly lower predicted probability of obstructive CAD than
those in Ontario. Overall, only 19.3% (95% CI, 18.7%-19.9%) of patients in New York were predicted
to have a greater than 50% probability of having obstructive CAD compared with 41.0% (95% CI,
40.6%-41.4%) in Ontario. At the lowest-risk category, when the predicted probability of obstructive
CAD was less than 15%, the proportion of patients in this category was 15.1% (95% CI, 14.6%-15.6%)
in New York and 6.9% (95% CI, 6.7%-7.1%) in Ontario. At the highest-risk spectrum, when the
predicted probability of obstructive CAD was greater than 75%, the proportion of patients was 1.4%
(95% CI, 1.2%-1.6%) in New York vs 7.9% (95% CI, 7.7%-8.1%) in Ontario.
In New York, predicted probabilities of obstructive CAD were 31.4% for hospitals that performed
cardiac catheterization, 32.8% for hospitals that performed cardiac catheterization and PCI, and
34.4% for full service hospitals. In Ontario, predicted probabilities of obstructive CAD were 44.6%
for hospitals that performed cardiac catheterization, 43.8% for hospitals that performed cardiac
catheterization and PCI, and 45.1% for full service hospitals
(P < .001).
We compared rates of coronary revascularization and mortality among patients who had undergone
cardiac catheterization between October 1, 2008, and December 31, 2010, which included 13 824
patients in New York and 40 794 patients in Ontario. Patients who had obstructive CAD in New
York were significantly more likely to undergo PCI and CABG surgery within 30 days after cardiac
catheterization than patients in Ontario (54.9% vs 34.6% for PCI, 20.4% vs 14.1% for CABG,
P < .001 for both comparisons). Mortality within 30 days of cardiac
catheterization was low in both New York and Ontario. At 30 days, crude mortality for patients
undergoing cardiac catheterization was slightly higher in New York at 0.65% (90 of 13 824; 95%
CI, 0.51%-0.78%) vs 0.38% (153 of 40 794; 95% CI, 0.32%-0.43%) in Ontario
(P < .001). However, this difference was driven primarily by higher
mortality for patients without obstructive CAD in New York at 0.62% (60 of 9709; 95% CI,
0.46%-0.77%) vs 0.27% (59 of 22 232; 95% CI, 0.20%-0.33%) in Ontario
(P < .001). There was no significant difference in New York and
Ontario in 30-day mortality among patients with obstructive CAD 0.73% (30 of 4115; 95% CI,
0.47%-0.99%) vs 0.51% (94 of 18 562; 95% CI, 0.40%-0.61%,
P = .08), or patients who received PCI at 0.29% (7 of 2393; 95% CI,
0.08%-0.51%) vs 0.25% (17 of 6698; 95% CI, 0.13%-0.37%, P = .75) or
CABG surgery at 0.67% (6 of 892; 95% CI, 0.14%-1.21%) vs 0.90% (24 of 2667; 95% CI, 0.54%-1.26%;
P = .52) .
We found that increased use of cardiac catheterization in New York relative to Ontario was
primarily the result of selecting more patients at low predicted probability of obstructive CAD. As
a result, the diagnostic yield (ie, the proportion of tested patients in whom disease was diagnosed)
of cardiac catheterization in New York was significantly lower than in Ontario. It is anticipated
that the cost of cardiac catheterization is higher in New York than Ontario; however, consistent
cost estimates of outpatient procedures are not widely available in New York. If we assumed all
cardiac catheterizations were performed on an outpatient basis at around $3000 per procedure as
estimated in Ontario19,20 and 30% of the
population undergoing cardiac catheterization had no prior cardiac disease, the hypothetical
scenario of New York’s adopting the same population rate of cardiac catheterization as in
Ontario (from 1185 per 100 000 to 605 per 100 000) could lead to potential savings of
approximately $75 million per year.
We have previously compared the market-oriented financing approach of New York with the
government-funded single-payer system of Ontario and found that New York has twice as many
interventional cardiologists, twice as many hospitals with cardiac invasive capabilities, and
accordingly performs almost exactly twice as many cardiac catheterizations per capita as
Ontario.11 This study represents an extension of the
previous work, which included detailed clinical characteristics and anatomical information of
patients undergoing cardiac catheterization. In addition to finding that New York patients
undergoing cardiac catheterization had a much lower predicted probability of having obstructive CAD,
several individual factors differed significantly between the 2 regions. First, the majority of
patients (58%) undergoing cardiac catheterization in New York did not have typical chest pain as
categorized by the Canadian Cardiovascular Society classification. Second, although the majority of
patients in both regions had noninvasive ischemic evaluation performed prior to cardiac
catheterization, only 5% of patients undergoing cardiac catheterization in New York vs 50% in
Ontario had high-risk findings on noninvasive stress testing. Although both data sets defined
high-risk findings in a similar manner, it is possible that the large discrepancy may be partly due
to systematic differences in the manner in which physicians interpret high-risk findings on
noninvasive stress testing in each region.
One of the primary reasons to perform cardiac catheterization is to detect patients with severe
CAD, for which coronary revascularization may improve clinical outcomes.21 We found that a more restrictive approach in selecting higher-risk patients for
cardiac catheterization was associated with improved detection of patients with single-vessel as
well as multivessel CAD. New York has historically performed twice as many cardiac catheterizations
as Ontario (1185 per 100 000 in New York vs 605 per 100 000 in Ontario).11 In this study, we observed that the detection rate for left
main or 3-vessel CAD suggests that New York was about half that of Ontario. Accordingly, the
estimated per capita detection rate was similar in both jurisdictions with the population rate of
left main stenosis of 29.6 per 100 000 in New York and 30.3 per 100 000 in Ontario, and
the population rate of 3-vessel disease estimated at 61.6 per 100 000 in New York and 59.3 per
100 000 in Ontario. These findings demonstrated that a more restrictive approach in selecting
patients for cardiac catheterization did not lead to substantial underdetection of patients with
surgical coronary anatomy on a per capita basis.
Several groups have proposed using obstructive CAD rate as a potential quality indicator to
enhance efficiency and improve quality.22,23 Our
study lends support to these proposals as we demonstrated the ability to increase diagnostic yield
of cardiac catheterization through improved patient selection. However, we do not believe the
current study can be used to determine the optimal rate of obstructive CAD or optimal selection
criteria for cardiac catheterization because decisions for procedure use are based on complex
interactions between patients, physicians, and the local environment.24 We do not wish to imply that the selection approach in Ontario is necessarily
We observed a higher rate of use for PCI and CABG at 30 days in New York than in Ontario among
patients with obstructive CAD. Mortality at 30 days among patients with obstructive CAD did not
differ significantly between the cohorts. These findings are consistent with prior regional
comparisons that suggest a region with much higher invasive capacity usually leads to greater use of
PCI procedures for more discretionary indications but may not reduce the frequency of adverse
cardiac outcomes.8- 10
Several potential limitations of this study deserve consideration. First, we compared
population-wide data from Ontario with a selected study sample of patients undergoing cardiac
catheterization in New York. To ensure generalizability of the New York cohort, we compared the
demographics and prevalence of risk factors among patients who underwent coronary revascularization
between our study sample and the entire New York, which demonstrated comparable results. Second,
although the availability of many clinical variables and a large study sample allowed us to compare
many characteristics associated with obstructive CAD, we were unable to assess the effect of race,
body mass index, or physician characteristics because these variables were not collected in both
databases. Finally, we were unable to formally apply appropriate-use criteria to compare the
suitability of cardiac catheterization because we lack data to create appropriateness scores in
Ontario.25 The appropriateness-use criteria have been
developed as 1 of the many existing tools designed to assist in improving quality of care and the
use of scarce health care resources. For patients with suspected CAD, the appropriate use criteria
categorized the appropriateness of procedure use based on the probability of CAD and most of the
patients at low risk of CAD are considered as having inappropriate indications for cardiac
catheterization. Similarly, we developed a prediction model to assess the probability of CAD and
found that New York selects patients at lower risk of having CAD, suggesting that New York may have
lower appropriateness scores than patients undergoing cardiac catheterization in Ontario. Although
we were unable to formally apply the appropriate-use criteria because of the lack of suitable data,
our study afforded similar insights by demonstrating that there is a significant opportunity to
improve patient care and improve the use of health care resources.
In conclusion, we found increased use of cardiac catheterization in New York compared with
Ontario and this reflects selection of patients at low risk of obstructive CAD. The observed pattern
of selecting patients with a higher probability of having coronary disease for cardiac
catheterization in Ontario was associated with improved diagnostic yield of the procedure.
Corresponding Author: Dennis T. Ko, MD, MSc, Institute for
Clinical Evaluative Sciences (ICES), G106-2075 Bayview Ave, Toronto, ON, Canada M4N 3M5 (firstname.lastname@example.org).
Author Contributions: Dr Ko had full access to
all of the data in the study and takes responsibility for the integrity of the data and the accuracy
of the data analysis.
Study concept and design: Ko, Tu, Samadashvili, Hannan.
Acquisition of data: Samadashvili, Hannan.
Analysis and interpretation of data: All authors.
Drafting of the manuscript: Ko.
Critical revision of the manuscript for important intellectual content: All
Statistical analysis: Samadashvili, Guo.
Obtained funding: Tu, Ko.
Administrative, technical, or material support: Hannan.
Study supervision: Tu, Hannan.
Conflict of Interest Disclosures: All authors
have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and
none were reported.
Funding/Support: This study was supported by
the Institute for Clinical Evaluative Sciences (ICES),
which is funded by an annual grant from the Ontario Ministry of Health and
Long-term Care (MOHLTC). The authors acknowledge that the clinical registry data
used in this publication are from the Cardiac Care Network of
Ontario and its member hospitals. The Cardiac Care Network of Ontario serves as a
system support to the MOHLTC, Local Health Integration Networks, and service providers and is
dedicated to improving the quality, efficiency, access and equity in the delivery of the continuum
of adult cardiovascular services in Ontario, Canada. The Cardiac Care Network of Ontario is funded
by the MOHLTC. The analysis of this study was supported by a Canadian
Institutes of Health Research (CIHR) operating grant MOP (102487). Dr Ko is
supported by a CIHR New Investigator Award. Dr Tu is supported by a Canada Research Chair in Health
Services Research and a Career Investigator Award from the Heart and
Stroke Foundation of Ontario. Dr Austin is supported by a Career Investigator Award
from the Heart and Stroke Foundation of Ontario.
Role of the Sponsor: The sponsors had no role in the design and conduct of the
study; collection, management, analysis, and interpretation of the data; preparation, review, or
approval of the manuscript; or decision to submit this manuscript.
Disclaimer: No endorsement by any of the supporting organizations is intended or
should be inferred.