- Email: [email protected]
Medical Therapy and Coronary Revascularization for Patients With Stable Coronary Artery Disease and Unclassified Appropriateness Score
Medical Therapy and Coronary Revascularization for Patients With Stable Coronary Artery Disease and Unclassified Appropriateness Score Mony Shuvy, MDa,b, Helen Guo, MScc, Harindra C. Wijeysundera, MD, PhDa,b,c,d, Christopher M. Feindel, MDb,e, Eric A. Cohen, MDa,b, Peter C. Austin, PhDb,c,d, Kori Kingsbury, MSNf, Madhu K. Natarajan, MD, MScg, Jack V. Tu, MD, PhDa,b,c,d, and Dennis T. Ko, MD, MSca,b,c,d,* Although the appropriate use criteria incorporate common clinical scenarios for coronary revascularization, a significant proportion of patients with stable coronary artery disease (CAD) cannot be assigned an appropriateness score. Our objective was to characterize these patients and to evaluate whether coronary revascularization is associated with improved outcomes. A population-based cohort of patients aged ‡66 years, who underwent cardiac catheterization in Ontario, Canada, were included. Clinical characteristics were compared between patients with and without an appropriateness score. Clinical outcomes between coronary revascularization and medical therapy in patients with unclassified appropriateness score were compared using the inverse probability of treatment-weighted propensity method for confounder adjustment. Of the 19,228 patients with stable CAD, 11.2% (2,153 patients) were not assigned to an appropriateness score, mostly (92.9%) because of a lack of ischemic evaluation or a noninterpretable test. These patients were older, had higher rate of severe angina, and had more medical co-morbidities compared to patients with an appropriateness score. The 2-year rate of death or myocardial infarction in patients with unclassified appropriateness score was 15.3% in the revascularization group versus 20.7% in the medical therapy group. After propensity weighting, revascularization was associated with significantly lower hazard ratio (0.70; 95% confidence interval 0.61 to 0.79) for death or myocardial infarction compared with medical therapy. In conclusion, in patients aged ‡66 years with stable CAD and unclassified appropriateness score, revascularization is associated with improved outcomes. Ó 2015 Elsevier Inc. All rights reserved. (Am J Cardiol 2015;116:1815e1821) Although the appropriate use criteria for coronary revascularization were constructed to incorporate most of the common clinical scenarios encountered in routine clinical practice, studies have demonstrated that as many as 25% of patients with stable coronary artery disease (CAD) cannot be assigned an appropriate score.1e5 Patients with stable CAD and unclassified appropriateness score are commonly encountered in clinical practice; however, little is known regarding their clinical characteristics, outcomes, and the impact of revascularization. Accordingly, the first objective of our study was to characterize patients who had unclassified a
Schulich Heart Centre, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; bUniversity of Toronto, Toronto, Ontario, Canada; cInstitute for Clinical Evaluative Sciences, Toronto, Ontario, Canada; dInstitute of Health Policy, Management and Evaluation, University of Toronto, Ontario, Canada; eUniversity Health Network, Toronto, Ontario, Canada; f Cardiac Care Network of Ontario, Toronto, Ontario, Canada; and gHamilton Health Sciences Centre, Toronto, Ontario, Canada. Manuscript received June 8, 2015; revised manuscript received and accepted September 19, 2015. Funding: The work was supported by Operating Grant 211089 from Canadian Institutes of Health Research. See page 1820 for disclosure information. *Corresponding author: Tel: (416) 480-4055 x 7835; fax: (416) 4806048. E-mail address: [email protected] (D.T. Ko). 0002-9149/15/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjcard.2015.09.020
appropriateness score for coronary revascularization. Second, we evaluated whether coronary revascularization is associated with improved clinical outcomes compared to medical therapy in patients with an unclassified appropriateness score. The Cardiac Care Network (CCN) registry, which maintains a prospective clinical registry of all patients undergoing cardiac invasive procedures in Ontario, Canada, was used to achieve our goals. Methods All hospitals in Ontario, Canada, are required to collect information prospectively on all cardiac catheterization, percutaneous coronary intervention (PCI), or coronary artery bypass grafting (CABG) procedures for the CCN registry. Demographics, clinical characteristics, and procedural information were gathered by clinical coordinators using procedure referral forms and hospital medical records. The registry was linked to the following population-based administrative databases in Ontario using unique encrypted patient health insurance numbers to protect patient confidentiality: the Canadian Institute for Health Information Discharge Abstract Database which captures all hospitalization records, the Ontario Health Insurance Plan database which includes information on physician service www.ajconline.org
1816
The American Journal of Cardiology (www.ajconline.org)
Table 1 Characteristics of patients with unclassified and assigned appropriateness score
Table 2 Characteristics of patients with unclassified appropriateness score after application of inverse probability treatment weights
Variable
Variable
Unclassified score (n¼2,153)
Age, mean SD (Years) 75.9 6.4 Women 735 (34.1%) CCS angina class No angina 610 (28.3%) I 226 (10.5%) II 497 (23.1%) III 508 (23.6%) IV 123 (5.7%) Unknown 189 (8.8%) Number of major epicardial arteries narrowed 0 6 (0.3%) 1 950 (44.1%) 2 720 (33.4%) 3 477 (22.2%) Left main or 3 vessel 495 (23.0%) Myocardial ischemia risk High 51 (2.4%) Low 15 (0.7%) Not done 1,547 (71.9%) Unknown 540 (25.1%) Left ventricular ejection fraction > 60% 665 (30.9%) 40%e59% 183 (8.5%) 20%e39% 105 (4.9%) 20% 57 (2.6%) Unknown 1,143 (53.1%) Cardiovascular risk factors Current smoker 346 (16.1%) Hypertension 1,628 (75.6%) Hyperlipidemia 1,543 (71.7%) Diabetes mellitus 715 (33.2%) Other conditions Previous myocardial 665 (30.9%) infarction Prior heart failure 402 (18.7%) Previous PCI 312 (14.5%) Cerebrovascular disease 227 (10.5%) Renal insufficiency 72 (3.3%) Peripheral vascular disease 218 (10.1%) Chronic obstructive 171 (7.9%) pulmonary disease Cardiac medications ACE inhibitors or ARBs 1,258 (58.4%) Statins 1,159 (53.8%) Beta blockers 1,150 (53.4%) Calcium channel blockers 750 (34.8%) Long acting nitrates 422 (19.6%) Clopidogrel 301 (14.0%) Physician and hospital characteristics Interventional cardiologist 1,469 (68.2%) Cardiac catheterization only 232 (10.8%) Cardiac catheterization and 1,539 (71.5%) PCI and CABG Cardiac catheterization, PCI 382 (17.7%) no CABG
Non-missing score (n¼17,074)
P value
74.7 5.8 <0.001 5,362 (31.4%) 0.01 <0.001 3,929 (23.0%) 2,275 (13.3%) 5,773 (33.8%) 4,451 (26.1%) 646 (3.8%) 0 236 7,660 5,353 3,825 5,345 7,670 4,425 4,858 121
8,197 2,196 897 258 5,526 2,614 13,335 13,212 5,675
(1.4%) <0.001 (44.9%) (31.4%) (22.4%) (31.3%) <0.001 (44.9%) (25.9%) (28.5%) (0.7%) <0.001 (48.0%) (12.9%) (5.3%) (1.5%) (32.4%) (15.3%) 0.357 (78.1%) 0.009 (77.4%) <0.001 (33.2%) 0.979
4,306 (25.2%) <0.001 1,810 2,346 1,600 355 1,571 1,145
(10.6%) <0.001 (13.7%) 0.341 (9.4%) 0.08 (2.1%) <0.001 (9.2%) 0.164 (6.7%) 0.032
10,522 10,018 10,058 6,299 3,217 2,290
(61.6%) 0.004 (58.7%) <0.001 (58.9%) <0.001 (36.9%) 0.062 (18.8%) 0.397 (13.4%) 0.467
11,762 (68.9%) 0.535 1,972 (11.5%) <0.001 13,114 (76.8%) 1,988 (11.6%)
ACE ¼ angiotensin converting enzyme; ARB ¼ angiotensin receptor blocker; CABG ¼ coronary artery bypass grafting; CCS ¼ Canadian Cardiovascular Society; PCI ¼ percutaneous coronary intervention.
Medical therapy (n¼855)
Revascularization Standardized (n¼1,298) difference
Age, mean SD (Years) 75.9 10.4 75.9 8.3 Women 288 (33.7%) 440 (33.9%) CCS angina class No angina 244 (28.5%) 373 (28.8%) I 89 (10.4%) 135 (10.4%) II 197 (23.1%) 293 (22.6%) III 208 (24.3%) 306 (23.6%) IV 45 (5.3%) 78 (6%) Unknown 72 (8.4%) 110 (8.5%) Number of major epicardial arteries narrowed 1 378 (44.2%) 578 (44.5%) 2 284 (33.2%) 430 (33.1%) 3 192 (22.4%) 287 (22.1%) Myocardial ischemia risk High 19 (2.2%) 30 (2.3%) Low 6 (0.7%) 8 (0.6%) Not done 616 (72.1%) 936 (72.1%) Unknown 215 (25.1%) 326 (25.1%) Left ventricular ejection fraction > 60% 265 (31%) 400 (30.8%) 40%e59% 70 (8.2%) 106 (8.2%) 20%e39% 41 (4.8%) 62 (4.8%) 20% 23 (2.7%) 41 (3.2%) Unknown 455 (53.2%) 687 (52.9%) Cardiovascular risk factors Current smoker 135 (15.8%) 208 (16%) Hypertension 651 (76.1%) 979 (75.4%) Hyperlipidemia 614 (71.8%) 932 (71.8%) Diabetes mellitus 286 (33.5%) 435 (33.5%) Other conditions Previous myocardial 262 (30.6%) 393 (30.3%) infarction Prior heart failure 159 (18.6%) 245 (18.9%) Previous percutaneous 127 (14.8%) 196 (15.1%) coronary intervention Cerebrovascular disease 91 (10.7%) 141 (10.9%) Renal insufficiency 28 (3.3%) 43 (3.3%) Peripheral vascular 88 (10.3%) 139 (10.7%) disease Chronic obstructive 68 (7.9%) 108 (8.3%) pulmonary disease Cardiovascular medications ACE inhibitors or ARBs 497 (58.1%) 758 (58.4%) Statins 460 (53.8%) 696 (53.6%) Beta blockers 453 (53%) 689 (53.1%) Calcium channel 308 (36%) 454 (35%) blockers Long acting nitrates 171 (20%) 257 (19.8%) Clopidogrel 124 (14.5%) 179 (13.8%) Physician and hospital characteristics Interventional 581 (68%) 881 (67.9%) cardiologist Cardiac catheterization 90 (10.5%) 141 (10.9%) only Cardiac catheterization 610 (71.4%) 928 (71.5%) and PCI and CABG Cardiac catheterization, 155 (18.1%) 228 (17.6%) PCI no CABG Please see Table 1 for abbreviations.
0.0117 0.0036 0.0084 0.0017 0.0113 0.0182 0.032 0.0029 0.0056 0.002 0.0073 0.0032 0.0103 0.001 0.001 0.0053 0.0016 0.0015 0.0313 0.006 0.0053 0.0164 0.0018 0.001 0.0053 0.0077 0.0074 0.0074 0.0018 0.014 0.0138
0.0045 0.0033 0.0027 0.0213 0.0045 0.0197 0.0026 0.0128 0.001 0.0114
Coronary Artery Disease/Revascularization and Unclassified Appropriateness Table 3 Clinical outcomes comparing medical therapy and revascularization Medical therapy (n¼855) Death (months) 1 6 12 18 24 Myocardial infarction (months) 1 6 12 18 24 Death or myocardial Infarction (months) 1 6 12 18 24
Revascularization (n¼1,298)
*P value
<0.001 26 80 115 135 155
(3%) (9.4%) (13.4%) (15.8%) (18.1%)
24 74 100 125 144
(1.9%) (5.7%) (7.7%) (9.6%) (11.1%) 0.5326
9 22 33 42 44
(1%) (2.6%) (3.9%) (4.9%) (5.2%)
32 49 65 67 74
(2.5%) (3.8%) (5%) (5.2%) (5.7%) <0.001
30 92 133 157 178
(3.5%) (10.8%) (15.6%) (18.4%) (20.8%)
49 112 147 173 199
(3.8%) (8.6%) (11.3%) (13.3%) (15.3%)
* p Value compared outcomes in the medically treated group vs the coronary revascularization group, obtained by the Kaplan-Meier curves in the propensity weighted sample.
claims, the Registered Persons Database which contains vital statistics and demographic information for all Ontarians, and the Ontario Drug Benefit database which includes prescription data for subjects aged 65 years. These linked data sets are described elsewhere and have been used extensively in clinical research.1,6,7 The need for informed consent was exempted under Ontario’s legislation regarding privacy of health information because participation in the CCN registry is mandatory. This study was approved by the research ethics board at the Sunnybrook Health Sciences Center. All patients aged 66 years who underwent cardiac catheterization from October 1, 2008, to July 31, 2011, in Ontario, Canada, were eligible for study inclusion. Patients with stable CAD were defined by excluding patients with acute coronary syndrome (unstable angina, myocardial infarction, and cardiogenic shock) in a manner consistent with previous research.6,8,9 Patients undergoing cardiac catheterization before scheduled cardiac operation were not included in the study. We selected patients with CAD, defined as luminal stenosis >50%, because the appropriateness criteria did not consider patients with normal to mild disease.2 For patients who had multiple cardiac catheterizations during the study period, only the first procedure was considered. An appropriateness score was assigned to each patient at the time of the cardiac catheterization according to the appropriate use criteria for coronary revascularization guideline, which was based on (1) clinical presentation and symptom severity; (2) intensity of anti-ischemic medical therapy; (3) ischemic burden as determined by noninvasive testing; and (4) extent of coronary artery stenosis on coronary
1817
angiography.1,2 Patients who could not be assigned an appropriateness score were considered to be in the unclassified appropriateness score category. The main outcome measure in the analysis comparing coronary revascularization and medical therapy was a composite measure of all-cause mortality or hospitalization for myocardial infarction. Mortality was determined using the Ontario Registered Persons Database. Myocardial infarction requiring hospitalization was identified from the Discharge Abstract Database using International Classification of Disease, Tenth Revision, code I21. To characterize patients with an unclassified appropriateness score, we compared the demographic and clinical characteristics of patients with and without an assigned appropriateness score. We used the chi-square tests for comparing categorical variables and the Wilcoxon ranksum test for continuous variables. We further compared patients with unclassified scores with patients who had appropriate score (7 to 9), uncertain score (4 to 6), and inappropriate score (1 to 3) for coronary revascularization. The potential benefit of coronary revascularization in patients with unclassified appropriateness score was evaluated by comparing outcomes of patients who underwent coronary revascularization with those who were treated medically. Patients who received PCI or CABG within 60 days after the index cardiac catheterization were categorized in the coronary revascularization group. To account for potential confounding between the treatment groups, we used the inverse probability of treatment-weighting method to account for systematic differences in baseline covariates between the revascularization and nonrevascularization groups. The propensity score, which was defined to be the probability of receiving coronary revascularization, was estimated with logistic regression analysis. The following characteristics which were selected on the basis of clinical knowledge were included as independent variables: age, gender, severity of angina, extent of coronary artery disease, ischemic evaluation, left ventricular ejection fraction, cardiac risk factors (smoking, hypertension, hyperlipidemia, diabetes), cardiac and medical co-morbidities (myocardial infarction, heart failure, chronic obstructive pulmonary disease, peripheral vascular disease, cerebrovascular disease, renal insufficiency, cancer, anemia, and trauma), cardiovascular medications, and physician (interventional cardiologist) and hospital characteristics (invasive capability). Patients were then weighted by the inverse of the probability of receiving the treatment that they actually received.10 The degree of balance between the treatment groups in the weighted cohort was assessed by computing the standardized difference, with differences of <0.1 indicating good balance.11 Adjusted KaplaneMeier survival curves were estimated for treated and untreated subjects separately in the weighted sample using a method described by Cole and Hernan.12 The effect of treatment on the hazard of clinical outcomes was then estimated using Cox proportional hazards model in which survival time was regressed on the recipient coronary revascularization. The inverse probability treatment weights were incorporated, and a robust variance estimator was used. We evaluated the potential benefit of the mode of coronary revascularization
1818
The American Journal of Cardiology (www.ajconline.org)
Figure 1. Clinical outcomes of coronary revascularization and medical therapy. KaplaneMeier curves comparing the outcomes of coronary revascularization and medical therapy for patients with unclassified appropriateness score are shown for different outcomes, (A) death; (B) death or myocardial infarction; and (C) myocardial infarction. Y axis shows proportions of patients free from events and x axis shows time from index coronary angiography in years. Inverse probability of treatment-weighted propensity-matched cohort of patients who underwent coronary revascularization are shown in black and patients who were treated medically are shown in grey.
on death or repeat hospitalization for myocardial infarction by the inverse probability of treatment-weighting method (i.e., we compared CABG vs medical therapy and PCI vs medical therapy). We also evaluated the potential impact of coronary revascularization on clinical outcomes in predefined subgroups (left main or 3-vessel CAD, 1- or
2-vessel CAD, no ischemic evaluation, uninterpretable ischemic evaluation) by balancing the covariates in each subgroup using the propensity-weighted method. Several additional analyses were performed to ensure our results were robust. First, we used multivariate proportional hazards models using coronary revascularization as
Coronary Artery Disease/Revascularization and Unclassified Appropriateness
1819
Table 4 Unadjusted and adjusted clinical outcomes comparing medical therapy and revascularization Unadjusted HR (95% CI)
Time dependent variable P value
HR (95% CI)
P value
IPTW analysis HR (95% CI)
P value
Death 0.51 (0.42, 0.62) <0.001 0.59 (0.48, 0.72) <0.001 0.59 (0.52, 0.68) <0.001 Myocardial infarction 1.25 (0.86, 1.83) 0.2483 0.89 (0.60, 1.33) 0.5722 1.08 (0.84, 1.39) 0.5326 Death or myocardial infarction 0.63 (0.53, 0.75) <0.001 0.66 (0.54, 0.80) <0.001 0.70 (0.61, 0.79) <0.001
IPTW with landmark analysis HR (95% CI)
P value
0.51 (0.47, 0.64) 0.76 (0.55, 1.04) 0.61 (0.53, 0.70)
<0.001 0.0889 <0.001
HR ¼ hazard ratio; IPTW ¼ inverse probability of treatment weight.
time-varying exposure to account for the potential difference in revascularization time after the index cardiac catheterization. The use of a time-dependent covariate enabled the treatment to vary over time during the observation period after cardiac catheterization and allowed us to estimate the effect of current exposure on the instantaneous hazard of adverse outcomes compared with medical therapy. Second, we repeated the propensity score analysis using a landmark analysis design. In doing so, we excluded patients who had died or were hospitalized for myocardial infarction within 60 days of the cardiac catheterization and used the inverse probability of treatment-weighting method for adjustment. For the landmark analysis, follow-up for outcomes began at 60 days after catheterization. All 2-sided p values <0.05 were considered statistically significant. SAS version 9.3 (SAS Institute, Cary, North Carolina) was used to conduct the analyses. Results After applying inclusion and exclusion criteria, our study cohort included 19,227 patients with stable CAD. Of these patients, 11.2% (2,153 patients) were not assigned to an appropriateness score. The most common reason for not having an appropriateness score was lack of ischemia evaluation: 73.9% (1,547 patients) had no evaluation and 24.8% (540 patients) had no interpretable results, followed by the lack of information about Canadian Cardiovascular Society (CCS) angina severity (7.1%, 153 patients). The demographics and baseline characteristics of patients with unclassified appropriateness score and the classified group are provided in Table 1. Patients in the classified group were further stratified into inappropriate, uncertain, and appropriate categories in Supplementary Table 1. Of the 2,153 patients with unclassified appropriateness score, the mean age was 75.9 years, 34.1% were women, and 29.3% had class III or IV CCS angina (Table 1). Patients with unclassified appropriateness score were slightly older, had more patients experiencing class IV angina, and also had a higher rate of asymptomatic patients without angina compared with patients with classified score. They also had significantly more co-morbidities such as previous myocardial infarction, heart failure, cerebrovascular disease, renal insufficiency, and chronic obstructive pulmonary disease (Table 1). Despite their greater risk characteristics, they were significantly less likely to undergo assessment of left ventricular ejection fraction before cardiac catheterization and were less likely to be prescribed angiotensin receptor enzyme inhibitors or angiotensin receptor blockers, b blockers, and statin medications (Table 1).
In the unclassified appropriateness score group, coronary revascularization was performed in 60.3% (1,298 patients), and 39.7% (855 patients) were treated medically. Patients who received coronary revascularization were slightly younger, more likely to have class III or IV angina, and more likely to have cardiac catheterization performed by an interventional cardiologist and at a full-service facility (Supplementary Table 2). Table 2 reports the baseline characteristics of the study population after application of inverse probability weights. All demographics and clinical characteristics were well balanced after propensity weight with all variables having a standardized difference of <0.1. The median follow-up of the unclassified appropriateness score study cohort was 3 years in the coronary revascularization group and 2.7 years in the medical therapy group. The rate of death or myocardial infarction was significantly lower in the coronary revascularization group compared to the medical therapy group (2-year rates 15.3% for revascularization vs 20.8% for medical therapy, p <0.001; Table 3). Difference in outcomes was driven mainly by a difference in death as the rate of myocardial infarction did not differ significantly. Clinical outcomes between the coronary revascularization and medical therapy groups began to diverge a month after the cardiac catheterization and continued to diverge over time (Figure 1). In subgroup analysis, patients who received CABG had significant improvement in the hazard of death or myocardial infarction (hazard ratio [HR] 0.64, 95% confidence interval [CI] 0.55 to 0.75) and those who received PCI (HR 0.85, 95% CI 0.74 to 0.98). Coronary revascularization was associated with a significantly lower HR (0.41, 95% CI 0.33 to 0.51, p <0.001) compared with medical therapy in patients with left main or 3-vessel CAD. In contrast, the HR (0.87; 95% CI 0.75 to 1.01, p ¼ 0.06) associated with coronary revascularization for patients who had 1 or 2 vessel disease did not reach statistical significance. There was no significant interaction between the hazard estimates between patients who had no stress test or uninterpretable tests. Hazard ratios comparing coronary revascularization with medical therapy were estimated using different statistical methods: (1) no confounding adjustment, (2) adjusted for revascularization as a time-dependent variable, (3) inverse probability weights propensity adjustment, and (4) inverse probability weights propensity adjustment with landmark analysis (Table 4). All methods showed similar overall outcomes, in which coronary revascularization was associated with a significant reduction in death, and death or myocardial infarction compared with medical therapy. The estimates using the inverse probability weights propensity adjustment analysis were the most conservative compared
1820
The American Journal of Cardiology (www.ajconline.org)
with other methods. For example, the HRs associated with coronary revascularization was 0.63 in the unadjusted proportion hazard model, 0.66 in time-dependent analysis, and 0.70 in propensity-weighting analysis. Discussion In this population-based cohort of older patients undergoing cardiac catheterization for stable CAD, we found that approximately 1 in 9 patients had unclassified appropriateness score. The most common reason why assignment of an appropriateness score was not possible was a lack of ischemic testing before cardiac catheterization. Patients with unclassified appropriateness score had significantly higher rates of previous cardiac conditions and medical co-morbidities compared to patients who had assigned scores. For patients with unclassified appropriateness score and obstructive CAD, coronary revascularization was associated with a 30% reduction in the hazard of death or myocardial infarction compared to medical therapy. Accordingly, patients with unclassified appropriateness score may not be equivalent as having inappropriate indications for revascularization. We found that 11.2% of patients were not able to be assigned to an appropriateness score in our study. In New York State, Hannan et al5 evaluated a similar cohort of patients with PCI that excluded acute coronary syndrome and previous CABG and found that an appropriateness score could not be assigned at a much higher rate of 28%. Bradley et al3 found that about 25% of patients in Washington State were unclassifiable. It is important to note that our study sample differed substantially than patients undergoing elective cardiac catheterization in the United States. In our cohort, 30% had CCS class III or IV angina and 23% were found to have left main or 3-vessel CAD. In contrast, Abdallah et al13 examined patients undergoing cardiac catheterization without previous stress testing in the United States and found that only about 15% of patients have severe angina and 11% had severe CAD. We found that patients who had unclassified appropriateness score were older and had significantly greater burden of co-morbidity than those who had assigned scores. Despite their greater risk characteristics, they were less likely to have undergone ischemia evaluation, assessed for left ventricular ejection fraction, and prescribed cardiac medications. Indeed, study has showed that patients who did not undergo stress testing before PCI were found to have a 13% greater risk of death compared with those who received preprocedural stress testing.14 It has been hypothesized that ischemic evaluation before PCI may lead to better clinical decision-making and therefore better outcomes. These studies were conducted using administrative data and lacked clinical detail on clinical factors and processes of care.14 Our study provides an alternative explanation that patients without ischemic evaluation have greater symptom burden or co-morbidity risk, were less likely to receive medical therapy, and are at greater risk of adverse outcome independent of the PCI procedure. Although the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial demonstrated that patients with stable CAD did not derive improved clinical outcomes with PCI compared to optimal
medical therapy, their findings may not be fully generalizable to our study population.15 First, patients in our study were at significantly greater risk compared with the COURAGE cohort that the observed mortality rate in the medical therapy at 9.3% at 6 months already exceeded that of COURAGE at the end of follow-up (8.3% at a median follow-up of almost 5 years).15 Second, 40% of the patients in the coronary revascularization group received CABG, whereas COURAGE evaluated primarily patients who underwent PCI procedures. Indeed, our subgroup analyses demonstrated that the benefits associated with coronary revascularization were largest in patients who had left main stenosis or 3-vessel CAD.15 Several aspects of this study merit consideration. First, observational studies are subject to the potential influence of confounding when comparing different treatments. Accordingly, we used a propensity method and successfully balanced all the observed patient and systematic factors. Nevertheless, these methods are still subjected to the potential influence of unmeasured confounding. Second, our study may be subjected to “survivorship bias,” and therefore, we conducted additional analysis using time varying exposure and landmark analysis and found that these methods yielded similar estimates. Third, reasons why patients do not first undergo ischemic evaluation are not completely understood. We hypothesized that it is likely multifactorial including clinical status with severe ischemic symptoms, comorbidities precluding the use of routine stress test, and inadequate access to noninvasive tests. Finally, although we found improved clinical outcome for coronary revascularization in patients who had unclassified appropriateness score, this finding should not be used as an endorsement for clinicians to forgo the use of ischemic evaluation before coronary revascularization. Acknowledgment: 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 an advisory body to the Ontario Ministry of Health and Long-Term Care (MOHLTC) and is dedicated to improving the quality, efficiency, access, and equity of adult cardiovascular services in Ontario, Canada. The Cardiac Care Network of Ontario and the Institute for Clinical Evaluative Sciences are funded by the MOHLTC. The results and conclusions are those of the authors and should not be attributed to any of the funding or participating organizations. The design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript were done independent of the funding agencies. Disclosures Dr. Wijeysundera is supported by a Distinguished Clinical Scientist Award from the Heart and Stroke Foundation (HSF) of Canada. Dr. Austin is supported by a Career Investigator Award from the HSF of Ontario. Dr. Tu is supported by a Canada Research Chair in Health Services Research and a Career Investigator Award from the HSF of
Coronary Artery Disease/Revascularization and Unclassified Appropriateness
Ontario. Dr. Ko is supported by a Clinician Investigator Award from the HSF of Ontario. Dr. Cohen has received financial support from Abbott Vascular, Medtronic, and Boston Scientific. The other authors have no financial conflicts of interest to declare. Supplementary Data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. amjcard.2015.09.020. 1. Ko DT, Chiu M, Guo H, Austin PC, Goeree R, Cohen E, Labinaz M, Tu JV. Safety and effectiveness of drug-eluting and bare-metal stents for patients with off- and on-label indications. J Am Coll Cardiol 2009;53:1773e1782. 2. Patel MR, Dehmer GJ, Hirshfeld JW, Smith PK, Spertus JA. ACCF/ SCAI/STS/AATS/AHA/ASNC/HFSA/SCCT 2012 Appropriate Use Criteria for Coronary Revascularization Focused Update. A Report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, Society for Cardiovascular Angiography and Interventions, Society of Thoracic Surgeons, American Association for Thoracic Surgery, American Heart Association, American Society of Nuclear Cardiology, and the Society of Cardiovascular Computed Tomography. J Am Coll Cardiol 2012;59:857e881. 3. Bradley SM, Maynard C, Bryson CL. Appropriateness of percutaneous coronary interventions in Washington state. Circ Cardiovasc Qual Outcomes 2012;5:445e453. 4. Chan PS, Patel MR, Klein LW, Krone RJ, Dehmer GJ, Kennedy K, Nallamothu BK, Weaver WD, Masoudi FA, Rumsfeld JS, Brindis RG, Spertus JA. Appropriateness of percutaneous coronary intervention. JAMA 2011;306:53e61. 5. Hannan EL, Cozzens K, Samadashvili Z, Walford G, Jacobs AK, Holmes DR Jr, Stamato NJ, Sharma S, Venditti FJ, Fergus I, King SB Iii. Appropriateness of coronary revascularization for patients without acute coronary syndromes. J Am Coll Cardiol 2012;59: 1870e1876.
1821
6. 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;310:163e169. 7. Tu JV, Bowen J, Chiu M, Ko DT, Austin PC, He Y, Hopkins R, Tarride JE, Blackhouse G, Lazzam C, Cohen EA, Goeree R. Effectiveness and safety of drug-eluting stents in Ontario. N Engl J Med 2007;357:1393e1402. 8. Patel MR, Peterson ED, Dai D, Brennan JM, Redberg RF, Anderson HV, Brindis RG, Douglas PS. Low diagnostic yield of elective coronary angiography. N Engl J Med 2010;362:886e895. 9. Wijeysundera HC, Bennell MC, Qiu F, Ko DT, Tu JV, Wijeysundera DN, Austin PC. Comparative-effectiveness of revascularization versus routine medical therapy for stable ischemic heart disease: a populationbased study. J Gen Intern Med 2014;29:1031e1039. 10. Austin PC. A tutorial and case study in propensity score analysis: an application to estimating the effect of in-hospital smoking cessation counseling on mortality. Multivariate Behav Res 2011;46:119e151. 11. Austin PC, Schuster T. The performance of different propensity score methods for estimating absolute effects of treatments on survival outcomes: a simulation study. Stat Methods Med Res 2014. http://dx.doi. org/10.1177/0962280213519716. 12. Cole SR, Hernan MA. Adjusted survival curves with inverse probability weights. Comput Methods Programs Biomed 2004;75: 45e49. 13. Abdallah MS, Spertus JA, Nallamothu BK, Kennedy KF, Arnold SV, Chan PS. Clinical symptoms and angiographic findings of patients undergoing elective coronary angiography without prior stress testing. Am J Cardiol 2014;114:348e354. 14. Lin GA, Lucas FL, Malenka DJ, Skinner J, Redberg RF. Mortality in Medicare patients undergoing elective percutaneous coronary intervention with or without antecedent stress testing. Circ Cardiovasc Qual Outcomes 2013;6:309e314. 15. Boden WE, O’Rourke RA, Teo KK, Hartigan PM, Maron DJ, Kostuk WJ, Knudtson M, Dada M, Casperson P, Harris CL, Chaitman BR, Shaw L, Gosselin G, Nawaz S, Title LM, Gau G, Blaustein AS, Booth DC, Bates ER, Spertus JA, Berman DS, Mancini GB, Weintraub WS; COURAGE Trial Research Group. Optimal medical therapy with or without PCI for stable coronary disease. N Engl J Med 2007;356: 1503e1516.
Schulich Heart Centre, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; bUniversity of Toronto, Toronto, Ontario, Canada; cInstitute for Clinical Evaluative Sciences, Toronto, Ontario, Canada; dInstitute of Health Policy, Management and Evaluation, University of Toronto, Ontario, Canada; eUniversity Health Network, Toronto, Ontario, Canada; f Cardiac Care Network of Ontario, Toronto, Ontario, Canada; and gHamilton Health Sciences Centre, Toronto, Ontario, Canada. Manuscript received June 8, 2015; revised manuscript received and accepted September 19, 2015. Funding: The work was supported by Operating Grant 211089 from Canadian Institutes of Health Research. See page 1820 for disclosure information. *Corresponding author: Tel: (416) 480-4055 x 7835; fax: (416) 4806048. E-mail address: [email protected] (D.T. Ko). 0002-9149/15/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.amjcard.2015.09.020
appropriateness score for coronary revascularization. Second, we evaluated whether coronary revascularization is associated with improved clinical outcomes compared to medical therapy in patients with an unclassified appropriateness score. The Cardiac Care Network (CCN) registry, which maintains a prospective clinical registry of all patients undergoing cardiac invasive procedures in Ontario, Canada, was used to achieve our goals. Methods All hospitals in Ontario, Canada, are required to collect information prospectively on all cardiac catheterization, percutaneous coronary intervention (PCI), or coronary artery bypass grafting (CABG) procedures for the CCN registry. Demographics, clinical characteristics, and procedural information were gathered by clinical coordinators using procedure referral forms and hospital medical records. The registry was linked to the following population-based administrative databases in Ontario using unique encrypted patient health insurance numbers to protect patient confidentiality: the Canadian Institute for Health Information Discharge Abstract Database which captures all hospitalization records, the Ontario Health Insurance Plan database which includes information on physician service www.ajconline.org
1816
The American Journal of Cardiology (www.ajconline.org)
Table 1 Characteristics of patients with unclassified and assigned appropriateness score
Table 2 Characteristics of patients with unclassified appropriateness score after application of inverse probability treatment weights
Variable
Variable
Unclassified score (n¼2,153)
Age, mean SD (Years) 75.9 6.4 Women 735 (34.1%) CCS angina class No angina 610 (28.3%) I 226 (10.5%) II 497 (23.1%) III 508 (23.6%) IV 123 (5.7%) Unknown 189 (8.8%) Number of major epicardial arteries narrowed 0 6 (0.3%) 1 950 (44.1%) 2 720 (33.4%) 3 477 (22.2%) Left main or 3 vessel 495 (23.0%) Myocardial ischemia risk High 51 (2.4%) Low 15 (0.7%) Not done 1,547 (71.9%) Unknown 540 (25.1%) Left ventricular ejection fraction > 60% 665 (30.9%) 40%e59% 183 (8.5%) 20%e39% 105 (4.9%) 20% 57 (2.6%) Unknown 1,143 (53.1%) Cardiovascular risk factors Current smoker 346 (16.1%) Hypertension 1,628 (75.6%) Hyperlipidemia 1,543 (71.7%) Diabetes mellitus 715 (33.2%) Other conditions Previous myocardial 665 (30.9%) infarction Prior heart failure 402 (18.7%) Previous PCI 312 (14.5%) Cerebrovascular disease 227 (10.5%) Renal insufficiency 72 (3.3%) Peripheral vascular disease 218 (10.1%) Chronic obstructive 171 (7.9%) pulmonary disease Cardiac medications ACE inhibitors or ARBs 1,258 (58.4%) Statins 1,159 (53.8%) Beta blockers 1,150 (53.4%) Calcium channel blockers 750 (34.8%) Long acting nitrates 422 (19.6%) Clopidogrel 301 (14.0%) Physician and hospital characteristics Interventional cardiologist 1,469 (68.2%) Cardiac catheterization only 232 (10.8%) Cardiac catheterization and 1,539 (71.5%) PCI and CABG Cardiac catheterization, PCI 382 (17.7%) no CABG
Non-missing score (n¼17,074)
P value
74.7 5.8 <0.001 5,362 (31.4%) 0.01 <0.001 3,929 (23.0%) 2,275 (13.3%) 5,773 (33.8%) 4,451 (26.1%) 646 (3.8%) 0 236 7,660 5,353 3,825 5,345 7,670 4,425 4,858 121
8,197 2,196 897 258 5,526 2,614 13,335 13,212 5,675
(1.4%) <0.001 (44.9%) (31.4%) (22.4%) (31.3%) <0.001 (44.9%) (25.9%) (28.5%) (0.7%) <0.001 (48.0%) (12.9%) (5.3%) (1.5%) (32.4%) (15.3%) 0.357 (78.1%) 0.009 (77.4%) <0.001 (33.2%) 0.979
4,306 (25.2%) <0.001 1,810 2,346 1,600 355 1,571 1,145
(10.6%) <0.001 (13.7%) 0.341 (9.4%) 0.08 (2.1%) <0.001 (9.2%) 0.164 (6.7%) 0.032
10,522 10,018 10,058 6,299 3,217 2,290
(61.6%) 0.004 (58.7%) <0.001 (58.9%) <0.001 (36.9%) 0.062 (18.8%) 0.397 (13.4%) 0.467
11,762 (68.9%) 0.535 1,972 (11.5%) <0.001 13,114 (76.8%) 1,988 (11.6%)
ACE ¼ angiotensin converting enzyme; ARB ¼ angiotensin receptor blocker; CABG ¼ coronary artery bypass grafting; CCS ¼ Canadian Cardiovascular Society; PCI ¼ percutaneous coronary intervention.
Medical therapy (n¼855)
Revascularization Standardized (n¼1,298) difference
Age, mean SD (Years) 75.9 10.4 75.9 8.3 Women 288 (33.7%) 440 (33.9%) CCS angina class No angina 244 (28.5%) 373 (28.8%) I 89 (10.4%) 135 (10.4%) II 197 (23.1%) 293 (22.6%) III 208 (24.3%) 306 (23.6%) IV 45 (5.3%) 78 (6%) Unknown 72 (8.4%) 110 (8.5%) Number of major epicardial arteries narrowed 1 378 (44.2%) 578 (44.5%) 2 284 (33.2%) 430 (33.1%) 3 192 (22.4%) 287 (22.1%) Myocardial ischemia risk High 19 (2.2%) 30 (2.3%) Low 6 (0.7%) 8 (0.6%) Not done 616 (72.1%) 936 (72.1%) Unknown 215 (25.1%) 326 (25.1%) Left ventricular ejection fraction > 60% 265 (31%) 400 (30.8%) 40%e59% 70 (8.2%) 106 (8.2%) 20%e39% 41 (4.8%) 62 (4.8%) 20% 23 (2.7%) 41 (3.2%) Unknown 455 (53.2%) 687 (52.9%) Cardiovascular risk factors Current smoker 135 (15.8%) 208 (16%) Hypertension 651 (76.1%) 979 (75.4%) Hyperlipidemia 614 (71.8%) 932 (71.8%) Diabetes mellitus 286 (33.5%) 435 (33.5%) Other conditions Previous myocardial 262 (30.6%) 393 (30.3%) infarction Prior heart failure 159 (18.6%) 245 (18.9%) Previous percutaneous 127 (14.8%) 196 (15.1%) coronary intervention Cerebrovascular disease 91 (10.7%) 141 (10.9%) Renal insufficiency 28 (3.3%) 43 (3.3%) Peripheral vascular 88 (10.3%) 139 (10.7%) disease Chronic obstructive 68 (7.9%) 108 (8.3%) pulmonary disease Cardiovascular medications ACE inhibitors or ARBs 497 (58.1%) 758 (58.4%) Statins 460 (53.8%) 696 (53.6%) Beta blockers 453 (53%) 689 (53.1%) Calcium channel 308 (36%) 454 (35%) blockers Long acting nitrates 171 (20%) 257 (19.8%) Clopidogrel 124 (14.5%) 179 (13.8%) Physician and hospital characteristics Interventional 581 (68%) 881 (67.9%) cardiologist Cardiac catheterization 90 (10.5%) 141 (10.9%) only Cardiac catheterization 610 (71.4%) 928 (71.5%) and PCI and CABG Cardiac catheterization, 155 (18.1%) 228 (17.6%) PCI no CABG Please see Table 1 for abbreviations.
0.0117 0.0036 0.0084 0.0017 0.0113 0.0182 0.032 0.0029 0.0056 0.002 0.0073 0.0032 0.0103 0.001 0.001 0.0053 0.0016 0.0015 0.0313 0.006 0.0053 0.0164 0.0018 0.001 0.0053 0.0077 0.0074 0.0074 0.0018 0.014 0.0138
0.0045 0.0033 0.0027 0.0213 0.0045 0.0197 0.0026 0.0128 0.001 0.0114
Coronary Artery Disease/Revascularization and Unclassified Appropriateness Table 3 Clinical outcomes comparing medical therapy and revascularization Medical therapy (n¼855) Death (months) 1 6 12 18 24 Myocardial infarction (months) 1 6 12 18 24 Death or myocardial Infarction (months) 1 6 12 18 24
Revascularization (n¼1,298)
*P value
<0.001 26 80 115 135 155
(3%) (9.4%) (13.4%) (15.8%) (18.1%)
24 74 100 125 144
(1.9%) (5.7%) (7.7%) (9.6%) (11.1%) 0.5326
9 22 33 42 44
(1%) (2.6%) (3.9%) (4.9%) (5.2%)
32 49 65 67 74
(2.5%) (3.8%) (5%) (5.2%) (5.7%) <0.001
30 92 133 157 178
(3.5%) (10.8%) (15.6%) (18.4%) (20.8%)
49 112 147 173 199
(3.8%) (8.6%) (11.3%) (13.3%) (15.3%)
* p Value compared outcomes in the medically treated group vs the coronary revascularization group, obtained by the Kaplan-Meier curves in the propensity weighted sample.
claims, the Registered Persons Database which contains vital statistics and demographic information for all Ontarians, and the Ontario Drug Benefit database which includes prescription data for subjects aged 65 years. These linked data sets are described elsewhere and have been used extensively in clinical research.1,6,7 The need for informed consent was exempted under Ontario’s legislation regarding privacy of health information because participation in the CCN registry is mandatory. This study was approved by the research ethics board at the Sunnybrook Health Sciences Center. All patients aged 66 years who underwent cardiac catheterization from October 1, 2008, to July 31, 2011, in Ontario, Canada, were eligible for study inclusion. Patients with stable CAD were defined by excluding patients with acute coronary syndrome (unstable angina, myocardial infarction, and cardiogenic shock) in a manner consistent with previous research.6,8,9 Patients undergoing cardiac catheterization before scheduled cardiac operation were not included in the study. We selected patients with CAD, defined as luminal stenosis >50%, because the appropriateness criteria did not consider patients with normal to mild disease.2 For patients who had multiple cardiac catheterizations during the study period, only the first procedure was considered. An appropriateness score was assigned to each patient at the time of the cardiac catheterization according to the appropriate use criteria for coronary revascularization guideline, which was based on (1) clinical presentation and symptom severity; (2) intensity of anti-ischemic medical therapy; (3) ischemic burden as determined by noninvasive testing; and (4) extent of coronary artery stenosis on coronary
1817
angiography.1,2 Patients who could not be assigned an appropriateness score were considered to be in the unclassified appropriateness score category. The main outcome measure in the analysis comparing coronary revascularization and medical therapy was a composite measure of all-cause mortality or hospitalization for myocardial infarction. Mortality was determined using the Ontario Registered Persons Database. Myocardial infarction requiring hospitalization was identified from the Discharge Abstract Database using International Classification of Disease, Tenth Revision, code I21. To characterize patients with an unclassified appropriateness score, we compared the demographic and clinical characteristics of patients with and without an assigned appropriateness score. We used the chi-square tests for comparing categorical variables and the Wilcoxon ranksum test for continuous variables. We further compared patients with unclassified scores with patients who had appropriate score (7 to 9), uncertain score (4 to 6), and inappropriate score (1 to 3) for coronary revascularization. The potential benefit of coronary revascularization in patients with unclassified appropriateness score was evaluated by comparing outcomes of patients who underwent coronary revascularization with those who were treated medically. Patients who received PCI or CABG within 60 days after the index cardiac catheterization were categorized in the coronary revascularization group. To account for potential confounding between the treatment groups, we used the inverse probability of treatment-weighting method to account for systematic differences in baseline covariates between the revascularization and nonrevascularization groups. The propensity score, which was defined to be the probability of receiving coronary revascularization, was estimated with logistic regression analysis. The following characteristics which were selected on the basis of clinical knowledge were included as independent variables: age, gender, severity of angina, extent of coronary artery disease, ischemic evaluation, left ventricular ejection fraction, cardiac risk factors (smoking, hypertension, hyperlipidemia, diabetes), cardiac and medical co-morbidities (myocardial infarction, heart failure, chronic obstructive pulmonary disease, peripheral vascular disease, cerebrovascular disease, renal insufficiency, cancer, anemia, and trauma), cardiovascular medications, and physician (interventional cardiologist) and hospital characteristics (invasive capability). Patients were then weighted by the inverse of the probability of receiving the treatment that they actually received.10 The degree of balance between the treatment groups in the weighted cohort was assessed by computing the standardized difference, with differences of <0.1 indicating good balance.11 Adjusted KaplaneMeier survival curves were estimated for treated and untreated subjects separately in the weighted sample using a method described by Cole and Hernan.12 The effect of treatment on the hazard of clinical outcomes was then estimated using Cox proportional hazards model in which survival time was regressed on the recipient coronary revascularization. The inverse probability treatment weights were incorporated, and a robust variance estimator was used. We evaluated the potential benefit of the mode of coronary revascularization
1818
The American Journal of Cardiology (www.ajconline.org)
Figure 1. Clinical outcomes of coronary revascularization and medical therapy. KaplaneMeier curves comparing the outcomes of coronary revascularization and medical therapy for patients with unclassified appropriateness score are shown for different outcomes, (A) death; (B) death or myocardial infarction; and (C) myocardial infarction. Y axis shows proportions of patients free from events and x axis shows time from index coronary angiography in years. Inverse probability of treatment-weighted propensity-matched cohort of patients who underwent coronary revascularization are shown in black and patients who were treated medically are shown in grey.
on death or repeat hospitalization for myocardial infarction by the inverse probability of treatment-weighting method (i.e., we compared CABG vs medical therapy and PCI vs medical therapy). We also evaluated the potential impact of coronary revascularization on clinical outcomes in predefined subgroups (left main or 3-vessel CAD, 1- or
2-vessel CAD, no ischemic evaluation, uninterpretable ischemic evaluation) by balancing the covariates in each subgroup using the propensity-weighted method. Several additional analyses were performed to ensure our results were robust. First, we used multivariate proportional hazards models using coronary revascularization as
Coronary Artery Disease/Revascularization and Unclassified Appropriateness
1819
Table 4 Unadjusted and adjusted clinical outcomes comparing medical therapy and revascularization Unadjusted HR (95% CI)
Time dependent variable P value
HR (95% CI)
P value
IPTW analysis HR (95% CI)
P value
Death 0.51 (0.42, 0.62) <0.001 0.59 (0.48, 0.72) <0.001 0.59 (0.52, 0.68) <0.001 Myocardial infarction 1.25 (0.86, 1.83) 0.2483 0.89 (0.60, 1.33) 0.5722 1.08 (0.84, 1.39) 0.5326 Death or myocardial infarction 0.63 (0.53, 0.75) <0.001 0.66 (0.54, 0.80) <0.001 0.70 (0.61, 0.79) <0.001
IPTW with landmark analysis HR (95% CI)
P value
0.51 (0.47, 0.64) 0.76 (0.55, 1.04) 0.61 (0.53, 0.70)
<0.001 0.0889 <0.001
HR ¼ hazard ratio; IPTW ¼ inverse probability of treatment weight.
time-varying exposure to account for the potential difference in revascularization time after the index cardiac catheterization. The use of a time-dependent covariate enabled the treatment to vary over time during the observation period after cardiac catheterization and allowed us to estimate the effect of current exposure on the instantaneous hazard of adverse outcomes compared with medical therapy. Second, we repeated the propensity score analysis using a landmark analysis design. In doing so, we excluded patients who had died or were hospitalized for myocardial infarction within 60 days of the cardiac catheterization and used the inverse probability of treatment-weighting method for adjustment. For the landmark analysis, follow-up for outcomes began at 60 days after catheterization. All 2-sided p values <0.05 were considered statistically significant. SAS version 9.3 (SAS Institute, Cary, North Carolina) was used to conduct the analyses. Results After applying inclusion and exclusion criteria, our study cohort included 19,227 patients with stable CAD. Of these patients, 11.2% (2,153 patients) were not assigned to an appropriateness score. The most common reason for not having an appropriateness score was lack of ischemia evaluation: 73.9% (1,547 patients) had no evaluation and 24.8% (540 patients) had no interpretable results, followed by the lack of information about Canadian Cardiovascular Society (CCS) angina severity (7.1%, 153 patients). The demographics and baseline characteristics of patients with unclassified appropriateness score and the classified group are provided in Table 1. Patients in the classified group were further stratified into inappropriate, uncertain, and appropriate categories in Supplementary Table 1. Of the 2,153 patients with unclassified appropriateness score, the mean age was 75.9 years, 34.1% were women, and 29.3% had class III or IV CCS angina (Table 1). Patients with unclassified appropriateness score were slightly older, had more patients experiencing class IV angina, and also had a higher rate of asymptomatic patients without angina compared with patients with classified score. They also had significantly more co-morbidities such as previous myocardial infarction, heart failure, cerebrovascular disease, renal insufficiency, and chronic obstructive pulmonary disease (Table 1). Despite their greater risk characteristics, they were significantly less likely to undergo assessment of left ventricular ejection fraction before cardiac catheterization and were less likely to be prescribed angiotensin receptor enzyme inhibitors or angiotensin receptor blockers, b blockers, and statin medications (Table 1).
In the unclassified appropriateness score group, coronary revascularization was performed in 60.3% (1,298 patients), and 39.7% (855 patients) were treated medically. Patients who received coronary revascularization were slightly younger, more likely to have class III or IV angina, and more likely to have cardiac catheterization performed by an interventional cardiologist and at a full-service facility (Supplementary Table 2). Table 2 reports the baseline characteristics of the study population after application of inverse probability weights. All demographics and clinical characteristics were well balanced after propensity weight with all variables having a standardized difference of <0.1. The median follow-up of the unclassified appropriateness score study cohort was 3 years in the coronary revascularization group and 2.7 years in the medical therapy group. The rate of death or myocardial infarction was significantly lower in the coronary revascularization group compared to the medical therapy group (2-year rates 15.3% for revascularization vs 20.8% for medical therapy, p <0.001; Table 3). Difference in outcomes was driven mainly by a difference in death as the rate of myocardial infarction did not differ significantly. Clinical outcomes between the coronary revascularization and medical therapy groups began to diverge a month after the cardiac catheterization and continued to diverge over time (Figure 1). In subgroup analysis, patients who received CABG had significant improvement in the hazard of death or myocardial infarction (hazard ratio [HR] 0.64, 95% confidence interval [CI] 0.55 to 0.75) and those who received PCI (HR 0.85, 95% CI 0.74 to 0.98). Coronary revascularization was associated with a significantly lower HR (0.41, 95% CI 0.33 to 0.51, p <0.001) compared with medical therapy in patients with left main or 3-vessel CAD. In contrast, the HR (0.87; 95% CI 0.75 to 1.01, p ¼ 0.06) associated with coronary revascularization for patients who had 1 or 2 vessel disease did not reach statistical significance. There was no significant interaction between the hazard estimates between patients who had no stress test or uninterpretable tests. Hazard ratios comparing coronary revascularization with medical therapy were estimated using different statistical methods: (1) no confounding adjustment, (2) adjusted for revascularization as a time-dependent variable, (3) inverse probability weights propensity adjustment, and (4) inverse probability weights propensity adjustment with landmark analysis (Table 4). All methods showed similar overall outcomes, in which coronary revascularization was associated with a significant reduction in death, and death or myocardial infarction compared with medical therapy. The estimates using the inverse probability weights propensity adjustment analysis were the most conservative compared
1820
The American Journal of Cardiology (www.ajconline.org)
with other methods. For example, the HRs associated with coronary revascularization was 0.63 in the unadjusted proportion hazard model, 0.66 in time-dependent analysis, and 0.70 in propensity-weighting analysis. Discussion In this population-based cohort of older patients undergoing cardiac catheterization for stable CAD, we found that approximately 1 in 9 patients had unclassified appropriateness score. The most common reason why assignment of an appropriateness score was not possible was a lack of ischemic testing before cardiac catheterization. Patients with unclassified appropriateness score had significantly higher rates of previous cardiac conditions and medical co-morbidities compared to patients who had assigned scores. For patients with unclassified appropriateness score and obstructive CAD, coronary revascularization was associated with a 30% reduction in the hazard of death or myocardial infarction compared to medical therapy. Accordingly, patients with unclassified appropriateness score may not be equivalent as having inappropriate indications for revascularization. We found that 11.2% of patients were not able to be assigned to an appropriateness score in our study. In New York State, Hannan et al5 evaluated a similar cohort of patients with PCI that excluded acute coronary syndrome and previous CABG and found that an appropriateness score could not be assigned at a much higher rate of 28%. Bradley et al3 found that about 25% of patients in Washington State were unclassifiable. It is important to note that our study sample differed substantially than patients undergoing elective cardiac catheterization in the United States. In our cohort, 30% had CCS class III or IV angina and 23% were found to have left main or 3-vessel CAD. In contrast, Abdallah et al13 examined patients undergoing cardiac catheterization without previous stress testing in the United States and found that only about 15% of patients have severe angina and 11% had severe CAD. We found that patients who had unclassified appropriateness score were older and had significantly greater burden of co-morbidity than those who had assigned scores. Despite their greater risk characteristics, they were less likely to have undergone ischemia evaluation, assessed for left ventricular ejection fraction, and prescribed cardiac medications. Indeed, study has showed that patients who did not undergo stress testing before PCI were found to have a 13% greater risk of death compared with those who received preprocedural stress testing.14 It has been hypothesized that ischemic evaluation before PCI may lead to better clinical decision-making and therefore better outcomes. These studies were conducted using administrative data and lacked clinical detail on clinical factors and processes of care.14 Our study provides an alternative explanation that patients without ischemic evaluation have greater symptom burden or co-morbidity risk, were less likely to receive medical therapy, and are at greater risk of adverse outcome independent of the PCI procedure. Although the Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial demonstrated that patients with stable CAD did not derive improved clinical outcomes with PCI compared to optimal
medical therapy, their findings may not be fully generalizable to our study population.15 First, patients in our study were at significantly greater risk compared with the COURAGE cohort that the observed mortality rate in the medical therapy at 9.3% at 6 months already exceeded that of COURAGE at the end of follow-up (8.3% at a median follow-up of almost 5 years).15 Second, 40% of the patients in the coronary revascularization group received CABG, whereas COURAGE evaluated primarily patients who underwent PCI procedures. Indeed, our subgroup analyses demonstrated that the benefits associated with coronary revascularization were largest in patients who had left main stenosis or 3-vessel CAD.15 Several aspects of this study merit consideration. First, observational studies are subject to the potential influence of confounding when comparing different treatments. Accordingly, we used a propensity method and successfully balanced all the observed patient and systematic factors. Nevertheless, these methods are still subjected to the potential influence of unmeasured confounding. Second, our study may be subjected to “survivorship bias,” and therefore, we conducted additional analysis using time varying exposure and landmark analysis and found that these methods yielded similar estimates. Third, reasons why patients do not first undergo ischemic evaluation are not completely understood. We hypothesized that it is likely multifactorial including clinical status with severe ischemic symptoms, comorbidities precluding the use of routine stress test, and inadequate access to noninvasive tests. Finally, although we found improved clinical outcome for coronary revascularization in patients who had unclassified appropriateness score, this finding should not be used as an endorsement for clinicians to forgo the use of ischemic evaluation before coronary revascularization. Acknowledgment: 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 an advisory body to the Ontario Ministry of Health and Long-Term Care (MOHLTC) and is dedicated to improving the quality, efficiency, access, and equity of adult cardiovascular services in Ontario, Canada. The Cardiac Care Network of Ontario and the Institute for Clinical Evaluative Sciences are funded by the MOHLTC. The results and conclusions are those of the authors and should not be attributed to any of the funding or participating organizations. The design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript were done independent of the funding agencies. Disclosures Dr. Wijeysundera is supported by a Distinguished Clinical Scientist Award from the Heart and Stroke Foundation (HSF) of Canada. Dr. Austin is supported by a Career Investigator Award from the HSF of Ontario. Dr. Tu is supported by a Canada Research Chair in Health Services Research and a Career Investigator Award from the HSF of
Coronary Artery Disease/Revascularization and Unclassified Appropriateness
Ontario. Dr. Ko is supported by a Clinician Investigator Award from the HSF of Ontario. Dr. Cohen has received financial support from Abbott Vascular, Medtronic, and Boston Scientific. The other authors have no financial conflicts of interest to declare. Supplementary Data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. amjcard.2015.09.020. 1. Ko DT, Chiu M, Guo H, Austin PC, Goeree R, Cohen E, Labinaz M, Tu JV. Safety and effectiveness of drug-eluting and bare-metal stents for patients with off- and on-label indications. J Am Coll Cardiol 2009;53:1773e1782. 2. Patel MR, Dehmer GJ, Hirshfeld JW, Smith PK, Spertus JA. ACCF/ SCAI/STS/AATS/AHA/ASNC/HFSA/SCCT 2012 Appropriate Use Criteria for Coronary Revascularization Focused Update. A Report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, Society for Cardiovascular Angiography and Interventions, Society of Thoracic Surgeons, American Association for Thoracic Surgery, American Heart Association, American Society of Nuclear Cardiology, and the Society of Cardiovascular Computed Tomography. J Am Coll Cardiol 2012;59:857e881. 3. Bradley SM, Maynard C, Bryson CL. Appropriateness of percutaneous coronary interventions in Washington state. Circ Cardiovasc Qual Outcomes 2012;5:445e453. 4. Chan PS, Patel MR, Klein LW, Krone RJ, Dehmer GJ, Kennedy K, Nallamothu BK, Weaver WD, Masoudi FA, Rumsfeld JS, Brindis RG, Spertus JA. Appropriateness of percutaneous coronary intervention. JAMA 2011;306:53e61. 5. Hannan EL, Cozzens K, Samadashvili Z, Walford G, Jacobs AK, Holmes DR Jr, Stamato NJ, Sharma S, Venditti FJ, Fergus I, King SB Iii. Appropriateness of coronary revascularization for patients without acute coronary syndromes. J Am Coll Cardiol 2012;59: 1870e1876.
1821
6. 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;310:163e169. 7. Tu JV, Bowen J, Chiu M, Ko DT, Austin PC, He Y, Hopkins R, Tarride JE, Blackhouse G, Lazzam C, Cohen EA, Goeree R. Effectiveness and safety of drug-eluting stents in Ontario. N Engl J Med 2007;357:1393e1402. 8. Patel MR, Peterson ED, Dai D, Brennan JM, Redberg RF, Anderson HV, Brindis RG, Douglas PS. Low diagnostic yield of elective coronary angiography. N Engl J Med 2010;362:886e895. 9. Wijeysundera HC, Bennell MC, Qiu F, Ko DT, Tu JV, Wijeysundera DN, Austin PC. Comparative-effectiveness of revascularization versus routine medical therapy for stable ischemic heart disease: a populationbased study. J Gen Intern Med 2014;29:1031e1039. 10. Austin PC. A tutorial and case study in propensity score analysis: an application to estimating the effect of in-hospital smoking cessation counseling on mortality. Multivariate Behav Res 2011;46:119e151. 11. Austin PC, Schuster T. The performance of different propensity score methods for estimating absolute effects of treatments on survival outcomes: a simulation study. Stat Methods Med Res 2014. http://dx.doi. org/10.1177/0962280213519716. 12. Cole SR, Hernan MA. Adjusted survival curves with inverse probability weights. Comput Methods Programs Biomed 2004;75: 45e49. 13. Abdallah MS, Spertus JA, Nallamothu BK, Kennedy KF, Arnold SV, Chan PS. Clinical symptoms and angiographic findings of patients undergoing elective coronary angiography without prior stress testing. Am J Cardiol 2014;114:348e354. 14. Lin GA, Lucas FL, Malenka DJ, Skinner J, Redberg RF. Mortality in Medicare patients undergoing elective percutaneous coronary intervention with or without antecedent stress testing. Circ Cardiovasc Qual Outcomes 2013;6:309e314. 15. Boden WE, O’Rourke RA, Teo KK, Hartigan PM, Maron DJ, Kostuk WJ, Knudtson M, Dada M, Casperson P, Harris CL, Chaitman BR, Shaw L, Gosselin G, Nawaz S, Title LM, Gau G, Blaustein AS, Booth DC, Bates ER, Spertus JA, Berman DS, Mancini GB, Weintraub WS; COURAGE Trial Research Group. Optimal medical therapy with or without PCI for stable coronary disease. N Engl J Med 2007;356: 1503e1516.