Abstract
Objectives
To compare the discriminatory value of differing risk scores for predicting clinical outcomes following PCI in routine practice.
Background
Various risk scores predict outcomes after PCI. However, these scores consider markedly different factors, from purely anatomical (SYNTAX risk score [SRS]) to purely clinical (ACEF, modified ACEF [ACEFmod], NCDR), while other scores combine both elements (Clinical SYNTAX score [CSS], NY State Risk Score [NYSRS]).
Methods
Patients with triple vessel and/or LM disease with 12 month follow-up were studied from a single center PCI registry. Exclusion criteria included STEMI presentation, prior revascularization and shock. Clinical events at 12 months were compared to baseline risk scores, according to score tertiles and area under receiver-operating-characteristic curves (AUC).
Results
We identified 584 eligible patients (69.8±12.3yrs, 405 males). All scores were predictive of mortality, with the SRS being least predictive (AUC=0.66). The most accurate scores for mortality were the CSS and ACEF (AUC=0.76 for both: p=0.019 and 0.08 vs. SRS, respectively). For TLR, while the SRS trended toward being positively predictive (p=0.075), several scores trended towards a negative association, which reached significance for the NCDR (p=0.045). The SRS and CSS were the only scores predictive of MI (both p<0.05). No score was particularly accurate for predicting MACE (death+MI+TLR), with AUCs ranging from 0.53 (NCDR) to 0.63 (SRS).
Conclusions
Competing factors influence mortality, MI and TLR after PCI. An increasing burden of comorbidities is associated with mortality, whereas anatomical complexity predicts MI. By combining these outcomes to predict MACE, all scores show reduced utility.
Indexing words: risk score, atherosclerosis, coronary artery disease, stenting
INTRODUCTION
The ability to accurately predict risk, complications and outcomes of a procedure is fundamental to clinical medicine. For patients diagnosed with significant multi-vessel coronary artery disease (CAD) who require revascularization, the SYNTAX risk score (SRS) is widely used to triage patients to either percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG) surgery [1–5]. In particular, a SRS of > 32, or potentially even > 22, indicates that the complexity and burden of CAD in any given patient is such that CABG, as compared to PCI, is the preferred and most efficacious revascularization strategy [3,5–7]. However, the SRS exclusively considers the number, complexity and nature of coronary artery lesions [2,3] and does not take into account clinical factors or patient co-morbidities. A newly created SYNTAX II score that includes clinical parameters has been described [8], but its use has not been reported in clinical practice.
In contrast to the SRS, numerous other risk scores and statistical models have been developed to predict outcomes following PCI and CABG that also incorporate clinical patient characteristics [9–22]. Some of these, such as the New York State Risk Score (NYSRS) [18] and the Clinical Syntax Score (CSS) [21], take into account both coronary anatomy and clinical features. On the other hand, the ACEF [20], modified ACEF (ACEFmod) [21] and National Cardiovascular Data Registry (NCDR) [19] risk scores consider only clinical factors. This distinction is of potential importance as prior studies have shown that lesion-based scores may be inferior to clinically-based scores for predicting post-PCI mortality [23], while emerging data has raised the possibility that the accuracy of these differing scores may vary depending on the clinical outcome(s) being considered [21,24,25]. In addition, these risk scores were originally designed for different purposes, such as predicting in-hospital outcomes (NYSRS, NCDR, ACEF), longer-term outcomes (CSS), or for other reasons such as assessing anatomical complexity (SRS). However, while objective information is lacking, with the exception of the SRS it would seem that in clinical practice these scores are often applied interchangeably and their original purpose not considered.
The aim of this study was to compare the predictive and discriminatory ability of these 6 risk scores with respect to differing 12 month clinical outcomes in stable ‘all-comer’ patients with three-vessel and/or left main CAD undergoing PCI.
METHODS
Data source and clinical definitions
Patients included in this analysis were evaluated from a single center, institutional review-board approved prospective interventional cardiology registry. All patients undergoing PCI are entered into this prospective PCI registry within 24 hours of the index intervention. Standardized data elements are collected at baseline that include clinical characteristics, procedural details, laboratory data, subsequent in-hospital clinical events and other test results associated with the procedure. For both PCI and balloon angioplasty only (PTCA), procedural success is defined as final TIMI grade II or III flow with less than < 50% residual stenosis. For diagnosing peri-procedural myocardial infarction (MI), all patients had at least 1 troponin-I drawn at least 6 hours after PCI. If this was above the normal limit, a CK-MB was drawn. After discharge, subjects are routinely contacted and undergo 30-day and 12 month follow-up. Major adverse cardiac events (MACE) and mortality are ascertained via phone contact with the patient, family, or primary physician. In addition, hospital records and the social security death index are queried for MACE and mortality. The following definitions were used: MACE = death, MI, or target lesion revascularization (TLR); TLR = repeat revascularization of the target lesion; MI = coronary ischemic syndrome resulting in creatine kinase-MB fraction > 5× upper limit of normal for peri-procedural events, or > 3× upper limit of normal for events occurring after hospital discharge. Troponin was substituted for creatine kinase-MB fraction if the latter was not available.
For adjudication of events, our database management team includes a dedicated physician who screens for potential adverse events. Those that clearly fall into a defined category (death, CK-MB clearly meeting MI definition) are not considered to require further adjudication and are entered directly into the database under the appropriate category. Our PCI reporting system includes extensive details regarding restenotic lesions and revascularization. If clearly documented in the PCI report, then this physician also has the discretion to enter TLR events directly into the database. For borderline events or those requiring clarification, documentation is obtained as required and a senior interventionalist and/or non-interventional cardiologist, or at times several cardiologists, are sought to reach consensus adjudication for the event.
Study population, data extraction and handling
Our interventional registry was retrospectively accessed and data were extracted on patients who underwent PCI between January 2007 and September 2010 with left main and/or three-vessel CAD. Eligible patients were required to exhibit a significant stenosis (> 50%) in either the left main coronary artery, or in each of the major epicardial vessels or their major side branches. Exclusion criteria were: cardiogenic shock (unassisted systolic blood pressure < 80 mmHg and/or cardiac index < 2.2 L/min/m2), presentation with ST-segment elevation MI (STEMI), any prior PCI or CABG, prior cardiac transplant, history of cerebrovascular accident with approximately modified Rankin grade 3 or worse neurological disability (moderate handicap, symptoms that restrict lifestyle and prevent totally independent exercise) [26], chronic liver disease classified as Child-Pugh class B or C, thrombocytopenia with platelet count < 50 × 109/L, intolerance or contraindication to acetylsalicylic acid or thienopyridines, need for major surgery or major concomitant non-cardiac disease with life expectancy < 12 months.
A total of 667 patients were potentially eligible for inclusion, of which 83 (12.3%) were excluded due to either incomplete 12 month follow-up or inability to calculate all six risk scores. As deaths were captured via the social security death index, mortality was not a cause of patients being excluded from this study.
PCI procedure
All patients presenting to the catheterization laboratory routinely received 325 mg aspirin > 90 minutes prior to angiography. In clopidogrel naïve patients, clopidogrel 600 mg was administered ‘on-the-table’ immediately prior to stent implantation, while patients already taking clopidogrel received an additional 300 mg dose. PCI was performed according to contemporary best clinical practice and using bivalirudin anticoagulation. Technical decisions regarding which lesions were treated, bifurcations, thrombus, calcification, diffuse disease, complex anatomy, stenting of side branches, use of glycoprotein IIb/IIIa inhibitors and the decision to implant bare metal (BMS) versus drug eluting stents (DES) was left to the discretion of the operators. Glycoprotein IIb/IIa inhibitors were only administered as a bolus dose and no patient received an infusion of these medications. Clopidogrel naïve patients who did not receive glycoprotein IIb/IIIa inhibitors received a 1 hour infusion of bivalirudin at the conclusion of the PCI. Aspirin was prescribed at 81 – 162 mg/day indefinitely in all subjects, while clopidogrel was prescribed at 75mg daily for at least 12 months.
Risk Score Calculation
Risk scores were arbitrarily selected for analysis in this study based on clinical familiarity and usage, and with a deliberate mix of anatomically-based and clinically-based algorithms. The specific inclusion of the NYSRS was due to mandatory reporting based on this score in our region (New York State). The SRS was calculated for each patient by scoring all coronary lesions using the SRS algorithm as described previously [1,2]. This was performed by 2 experienced observers, blinded to the patient’s clinical outcome. If differing inter-observer grading occurred, consensus was reached after review. Other risk scores were calculated based on relevant clinical and coronary angiographic features and patient comorbidities (Supplementary Table I).
Statistical methods
We applied a standard test to assess distribution skewness and kurtosis of all variables [27], and unless otherwise stated data are summarized as mean ± standard deviation (SD) or, for data with a skewed distribution, as median and range. Patients were assigned into tertiles for each risk score, with tertiles being estimated from data presented here (and not using other tertiles that have been previously associated with these scores). Comparisons of means in two independent groups were performed by two-tail Student’s t test. For skewed data, the Mann–Whitney rank-sum test was used. Multiple group comparisons were performed using Pearson‧s chi-square test for categorical variables and ANOVA for continuous variables.
Risk-score discriminatory ability was also assessed using the area under the receiver-operating-characteristic (ROC) curve. The estimation of the area under the ROC curve (AUC) was performed by nonparametric ROC analysis and the hypothesis of equal areas under two ROC curves from the same sample was tested using the method described in DeLong et al [28]. Statistical analysis was carried out using Stata 11.0 software (StataCorp, College Station, TX).
RESULTS
We identified 584 patients with three-vessel and/or left main CAD who underwent PCI and who were eligible for inclusion. For each patient, we calculated the SRS, CSS, NYSRS, ACEF, ACEFmod and NCDR risk scores (Supplementary Table I). Patient demographics are presented in Table I. The majority of patients were male (405/584, 69.4%) and clinical features were generally typical of those with multivessel CAD.
Table I.
Baseline demographics and other patient characteristics. Data is presented as n (%), or mean ± SD. Abbreviations not previously defined: BMI, body mass index; COPD, chronic obstructive pulmonary disease; CCS, Canadian Cardiovascular Society; eGFR, estimated glomerular filtration rate (calculated using MDRD formula); NSTEMI, non ST-segment elevation myocardial infarction. For CCS, data were available for 569 patients.
| Females (n = 179) |
Males (n = 405) |
All Subjects (n = 584) |
P value females vs. males |
|
|---|---|---|---|---|
| Age (years) | 75.8 ± 10.6 | 67.2 ± 12.2 | 69.8 ± 12.3 | < 0.0001 |
| Weight (kg) | 70.3 ± 17.9 | 83.0 ± 18.0 | 79.1 ± 18.9 | < 0.0001 |
| BMI (kg.m−2) | 28.3 ± 6.6 | 28.2 ± 5.5 | 28.2 ± 5.9 | 0.79 |
| Comorbidities | ||||
| Prior Stroke | 20 (11.2%) | 39 (9.6%) | 59 (10.1%) | 0.57 |
| Current smoker | 22 (12.3%) | 106 (26.2%) | 128 (22.0%) | < 0.0001 |
| Hypertension | 170 (95.0%) | 370 (91.4%) | 540 (92.5%) | 0.13 |
| COPD | 27 (15.1%) | 33 (8.2%) | 60 (10.3%) | 0.011 |
| Diabetes | 88 (49.2%) | 151 (37.3%) | 239 (40.9%) | 0.007 |
| Laboratory Data | ||||
| Total cholesterol (mg/dL) | 156.8 ± 40.9 | 151.7 ± 43.4 | 153.3 ± 42.7 | 0.19 |
| Hemoglobin (g/dL) | 11.7 ± 1.8 | 13.4 ± 1.9 | 12.9 ± 2.0 | < 0.0001 |
| eGFR (ml/min/1.72m2) | 59.0 ± 29.5 | 80.5 ± 34.6 | 74.1 ± 34.6 | < 0.0001 |
| Cardiovascular Status | ||||
| CCS angina class | ||||
| 0 | 10 (5.9%) | 23 (5.8%) | 33 (5.8%) | 0.11 |
| I/II | 69 (40.8%) | 201 (50.3%) | 270 (47.5%) | |
| III/IV | 90 (53.3%) | 176 (44.0%) | 266 (46.8%) | |
| LVEF (%) | 53.3 ± 10.9 | 51.3 ±13.2 | 51.9 ± 12.6 | 0.080 |
| PAD | 20 (11.8%) | 44 (11.0%) | 64 (11.3%) | 0.77 |
Median, range and tertiles for risk scores are presented in Table II and distributions in Figure 1. The SRS was the only risk score for which females and males had similar scores; for all other scores females had significantly higher scores than males.
Table II.
Median, range and tertiles for risk scores.
| Tertiles of Risk Scores |
Females (n = 179) |
Males (n = 405) |
All Subjects n = 584) |
P value males v. females |
||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Low | Mid | High | Median | Range | Median | Range | Median | Range | ||
| SRS | < 18.0 | 18.0 – 25.5 | > 25.5 | 22.0 | 8.0 – 54.5 | 21.0 | 5.0 – 66.0 | 21.5 | 5.0 – 66.0 | 0.45 |
| CSS | < 26.9 | 26.9 – 60.0 | > 60.0 | 54.5 | 9.3 – 309.2 | 33.8 | 4.4 – 409.7 | 38.7 | 4.4 – 409.7 | < 0.0001 |
| NYSRS | < 5.0 | 5.0 – 9.0 | > 9.0 | 8.0 | 0.0 – 26.0 | 5.0 | 0.0 – 29.0 | 6.0 | 0.0 – 29.0 | < 0.0001 |
| ACEF | < 1.17 | 1.17 – 1.51 | > 1.51 | 1.38 | 0.82 – 4.30 | 1.25 | 0.55 – 9.10 | 1.31 | 0.55 – 9.10 | 0.0001 |
| ACEFmod | < 1.23 | 1.23 – 2.81 | > 2.81 | 2.60 | 0.85 – 7.50 | 1.40 | 0.55 – 13.10 | 1.93 | 0.55 – 13.10 | < 0.0001 |
| NCDR | < 15 | 15 – 25 | > 25 | 24.0 | 0.0 – 56.0 | 16.0 | 0.0 – 69.0 | 18.0 | 0.0 – 69.0 | < 0.0001 |
Figure 1.
Risk score distributions.
Procedural details are presented according to risk score tertiles in Table III, showing that the highest tertile SRS patients tended to had more lesions treated, more stents implanted and a greater total implanted stent length. The opposite was true for clinically-based risk scores, where patients in the highest tertiles generally had fewer lesions treated and lower total stent length. There was no particular trend identified for these measures and tertiles of the CSS. Regarding the use of BMS versus DES, across all risk scores there was increasing use of BMS with higher tertiles of risk, while for DES the opposite was observed, with all scores except the SRS showing progressively less DES use with increasing clinical risk (Table III). Revascularization was performed in a single PCI session for 40.7% of patients, in 2 sessions for 41.9% and staged across 3 sessions in 17.4%. Of 1869 total lesions treated (mean 3.2 ± 1.7 per patient), procedural success was achieved in 97.4% (1821/1869), while for 2.6% of lesions (48/1869) PCI was unsuccessful.
Table III.
Procedural details according to tertiles of risk scores (mean ± SD). Data shown is per-patient.
| Total Number of Lesions Treated |
Number of Lesions Treated by PTCA only |
Total Number of Implanted Stents |
Number Of BMS Implanted |
Number of DES Implanted |
Total Length of Stents Implanted |
|
|---|---|---|---|---|---|---|
| SRS | ||||||
| Low | 2.8 ± 1.4 | 0.03 ± 0.63 | 2.8 ± 1.5 | 0.6 ± 1.1 | 2.2 ± 1.7 | 55.1 ±35.4 |
| Intermed | 3.2 ± 1.7 | 0.2 ± 0.8 | 3.0 ± 1.7 | 0.6 ± 1.0 | 2.4 ± 2.0 | 63.5 ± 41.6 |
| High | 3.6 ± 1.9 | 0.3 ± 0.9 | 3.4 ± 1.9 | 1.0 ± 1.3 | 2.5 ± 2.2 | 70.9 ± 45.4 |
| P value | < 0.0001 | 0.01 | 0.0029 | 0.004 | 0.24 | 0.0008 |
| CSS | ||||||
| Low | 3.2 ± 1.6 | 0.1 ± 0.8 | 3.1 ± 1.6 | 0.5 ± 1.0 | 2.6 ± 1.7 | 65.4 ± 39.1 |
| Intermed | 3.4 ± 1.8 | 0.2 ± 0.8 | 3.2 ± 1.8 | 0.7 ± 1.0 | 2.6 ± 2.1 | 65.7 ± 43.3 |
| High | 3.0 ± 1.7 | 0.2 ± 0.8 | 2.9 ± 1.7 | 1.0 ± 1.3 | 1.8 ± 2.0 | 58.0 ± 41.2 |
| P value | 0.19 | 0.33 | 0.18 | < 0.0001 | < 0.0001 | 0.11 |
| NYSRS | ||||||
| Low | 3.5 ± 1.7 | 0.1 ± 0.8 | 3.4 ± 1.7 | 0.4 ± 0.9 | 3.0 ± 1.9 | 71.7 ± 43.0 |
| Intermed | 3.1 ± 1.8 | 0.1 ± 0.8 | 3.0 ± 1.8 | 0.7 ± 1.1 | 2.3 ± 1.9 | 61.3 ± 42.0 |
| High | 3.0 ± 1.6 | 0.2 ± 0.8 | 2.7 ± 1.6 | 1.1 ± 1.3 | 1.6 ± 1.8 | 53.7 ± 35.9 |
| P value | 0.0028 | 0.37 | 0.0007 | < 0.0001 | < 0.0001 | 0.0001 |
| ACEF | ||||||
| Low | 3.6 ± 1.8 | 0.1 ± 0.8 | 3.5 ± 1.8 | 0.5 ± 1.0 | 3.0 ± 2.0 | 74.0 ± 44.6 |
| Intermed | 3.1 ± 1.6 | 0.1 ± 0.8 | 3.0 ± 1.7 | 0.6 ± 1.1 | 2.4 ± 1.8 | 60.4 ± 37.0 |
| High | 2.9 ± 1.7 | 0.2 ± 0.8 | 2.7 ± 1.7 | 1.1 ± 1.3 | 1.6 ± 1.8 | 54.5 ± 39.7 |
| P value | 0.0001 | 0.47 | < 0.0001 | < 0.0001 | < 0.0001 | < 0.0001 |
| ACEFmod | ||||||
| Low | 3.5 ± 1.7 | 0.1 ± 0.9 | 3.4 ± 1.7 | 0.5 ± 1.1 | 2.8 ± 1.8 | 72.0 ± 41.5 |
| Intermed | 3.3 ± 1.9 | 0.2 ± 0.8 | 3.0 ± 1.9 | 0.6 ± 1.0 | 2.5 ± 2.0 | 62.6 ± 43.2 |
| High | 2.8 ± 1.5 | 0.1 ± 0.8 | 2.7 ± 1.5 | 1.0 ± 1.3 | 1.7 ± 1.8 | 54.3 ± 37.3 |
| P value | 0.0008 | 0.4 | 0.0013 | < 0.0001 | < 0.0001 | 0.0001 |
| NCDR | ||||||
| Low | 3.4 ± 1.8 | 0.2 ± 0.8 | 3.2 ± 1.7 | 0.6 ± 1.0 | 2.6 ± 1.9 | 67.0 ± 42.5 |
| Intermed | 3.4 ± 1.8 | 0.1 ± 0.8 | 3.4 ± 1.8 | 0.6 ± 1.1 | 2.8 ± 2.0 | 70.9 ± 43.1 |
| High | 2.8 ± 1.6 | 0.2 ± 0.7 | 2.6 ± 1.6 | 1.1 ± 1.3 | 1.5 ± 1.7 | 50.1 ± 34.6 |
| P value | 0.0002 | 0.21 | < 0.0001 | < 0.0001 | < 0.0001 | 0.006 |
During the 1 year study period 40 patients died, 25 suffered an MI (16 peri-procedural, 9 spontaneous) and 56 underwent TLR (Supplementary Table II). Risk scores were considered by tertiles and compared to the clinical outcomes of death, MI and TLR at 12 months (Table IV). Mortality was the only outcome predicted by all risk scores (Figure 2, Table IV). The SRS and CSS were the only scores predictive of MI (Figure 3, Table IV). For TLR, the SRS was the only risk score that showed a positive trend for this outcome (p = 0.075), however, several scores showed a negative trend for predicting TLR, which was significant for the NCDR score (p = 0.045) (Figure 4, Table IV). For MACE, the SRS, CSS, NYSRS and the ACEF were predictive, while the ACEFmod and NCDR showed no association (Figure 5, Table IV).
Table IV.
Predictive ability of risk scores for clinical events at 12 months according to tertiles of risk.
| Mortality | TLR | MI | MACE | |||||
|---|---|---|---|---|---|---|---|---|
| n | % | n | % | n | % | n | % | |
| SRS | ||||||||
| Low | 6 | 3.05 | 13 | 6.60 | 4 | 2.03 | 21 | 10.66 |
| Intermed | 13 | 6.50 | 18 | 9.00 | 7 | 3.50 | 36 | 18.00 |
| High | 21 | 11.23 | 25 | 13.37 | 14 | 7.49 | 52 | 27.81 |
| P value | 0.006 | 0.075 | 0.024 | < 0.0001 | ||||
| CSS | ||||||||
| Low | 6 | 3.08 | 20 | 10.26 | 3 | 1.54 | 29 | 14.87 |
| Intermed | 7 | 3.59 | 22 | 11.28 | 9 | 4.62 | 32 | 16.41 |
| High | 27 | 13.92 | 14 | 7.22 | 13 | 6.70 | 48 | 24.74 |
| P value | < 0.0001 | 0.367 | 0.041 | 0.027 | ||||
| NYSRS | ||||||||
| Low | 5 | 2.24 | 23 | 10.31 | 8 | 3.59 | 33 | 14.80 |
| Intermed | 7 | 3.72 | 16 | 8.51 | 6 | 3.19 | 26 | 13.83 |
| High | 28 | 16.18 | 17 | 9.83 | 11 | 6.36 | 50 | 28.90 |
| P value | < 0.0001 | 0.891 | 0.269 | < 0.0001 | ||||
| ACEF | ||||||||
| Low | 5 | 2.55 | 24 | 12.24 | 6 | 3.06 | 32 | 16.33 |
| Intermed | 6 | 3.09 | 14 | 7.22 | 7 | 3.61 | 25 | 12.89 |
| High | 29 | 14.95 | 18 | 9.28 | 12 | 6.19 | 52 | 26.80 |
| P value | < 0.0001 | 0.237 | 0.267 | 0.001 | ||||
| ACEFmod | ||||||||
| Low | 5 | 2.53 | 23 | 11.62 | 6 | 3.03 | 32 | 16.16 |
| Intermed | 10 | 5.21 | 21 | 10.94 | 7 | 3.65 | 35 | 18.23 |
| High | 25 | 12.89 | 12 | 6.19 | 12 | 6.19 | 42 | 21.65 |
| P value | < 0.0001 | 0.14 | 0.264 | 0.372 | ||||
| NCDR | ||||||||
| Low | 7 | 3.04 | 28 | 12.17 | 7 | 3.04 | 39 | 16.96 |
| Intermed | 12 | 6.78 | 19 | 10.73 | 11 | 6.21 | 36 | 20.34 |
| High | 21 | 11.86 | 9 | 5.08 | 7 | 3.95 | 34 | 19.21 |
| P value | 0.002 | 0.045 | 0.284 | 0.669 | ||||
Figure 2.
Predictive ability of risk scores for mortality by tertiles of risk.
Figure 3.
Predictive ability of risk scores for MI by tertiles of risk.
Figure 4.
Predictive ability of risk scores for TLR by tertiles of risk.
Figure 5.
Predictive ability of risk scores for MACE by tertiles of risk.
ROC curve and AUC analysis was performed to further explore the discriminatory ability of these scores for mortality, MI and MACE. ROC curves for TLR are not presented, as no score was positively predictive of TLR by tertile analysis. As the only score to consider purely anatomical coronary artery characteristics, the SRS was chosen as the reference score. For predicting mortality, the SRS had the lowest AUC, and was statistically inferior to the CSS score (AUC: SRS 0.6616, CSS 0.7550; p = 0.019 CSS vs. SRS) (Figure 6). The ACEF score also trended towards being superior to the SRS for predicting mortality (AUC: ACEF 0.7586; p = 0.082 ACEF vs. SRS). For predicting MI, the SRS had the highest AUC (0.6648), but was not statistically superior to any other risk score (Supplementary Figure I). For MACE, the SRS had the greatest AUC and was superior to the NCDR score (AUC: SRS 0.6282, NCDR 0.5333; p = 0.021 NCDR vs. SRS), with a trend towards being statistically superior to the ACEFmod score (AUC: ACEFmod 0.5589; p = 0.076 ACEFmod vs. SRS) (Figure 7).
Figure 6.
ROC curves and comparisons for mortality.
Figure 7.
ROC curves and comparisons for MACE.
DISCUSSION
While CABG remains the preferred method of revascularization for patients with complex CAD [5,29], for reasons such as strong patient preference or comorbidities PCI continues to be widely performed for this indication [30]. While overall outcomes of PCI appear satisfactory, an important component of any treatment modality is that patients can be prospectively managed and advised with respect to their risk for adverse outcomes. This study was designed to compare the utility and respective strengths and limitations of differing risk scores for predicting various clinical outcomes in contemporary PCI patients with triple vessel and/or left main CAD. We found that clinical factors and clinically-based risk scores are generally superior for predicting mortality, although the CSS (considering both clinical and angiographic factors) also performed well for this outcome. On the other hand, purely angiographic factors (SRS) are the most predictive of MI. Strikingly, while the SRS was the most accurate score for predicting MI, it was the weakest of all scores for predicting mortality. Most notably, for predicting MACE all scores showed relatively poor discriminatory ability. As a whole, these findings illustrate the limitations of applying a mathematical model to a complex clinical substrate, and raise several important additional issues regarding risk score prognostication.
Importance of comorbidities
Recently, much attention has been given to the importance of clinical comorbidities in cardiovascular risk scores [8,21,22]. As our data has confirmed, these clinical aspects are critical for predicting mortality after PCI. Of all the outcomes considered here, risk score performance was clearly superior for mortality, with the highest values for the area under the mortality ROC curves being in the order of 0.74 – 0.76 (ACEF, CSS, ACEFmod, NYSRS). However, our data has demonstrated that these clinical factors are of reduced utility for predicting MI and TLR. Indeed, the purely clinical NCDR risk score was a significant negative predictor of TLR. We suspect that this may be due to a number of reasons. Firstly, in ‘real world’ practice, patients with an extremely high burden of comorbidities may undergo less extensive PCI-based revascularization (they receive fewer stents and have less total stented length that is at risk for restenosis). In addition, patients with a high burden of clinical comorbidities may also die, or be conservatively managed in preference to returning to the catheterization laboratory.
Given these competing aspects, when outcomes are combined to derive MACE, no score was particularly effective. Most accurate was the SRS, however, the AUC for the SRS for MACE was only 0.6282. Some of the more poorly performing scores had an AUC for MACE that approached 0.5, indicting predictability equivalent to random chance.
Importance of patient selection
An important aspect of our study is the patient population. Patients presented here represent ‘real world’ clinical practice. In many cases, those with a high SRS that underwent PCI in this study were unwilling to undergo CABG or were deemed at excessive risk for surgery. It is in this precise group of high risk patients that accurate assessment of risk should be considered the most crucial. In contrast, several of the risk scores evaluated here were developed or validated in patients that were at significantly lower risk for adverse events and/or were participating in clinical trials. For example, patients used to derive the CSS from the ARTS-II study were classified by tertiles as CSSlow ≤ 15.6, CSSmid 15.6 – 27.5 and CSShigh > 27.5, with the highest CSS being 209 [21]. In an ARTS-II substudy considering only patients with triple vessel CAD, the CSShigh tertile was defined as > 31.2 [21]. In contrast, our tertiles for the CSS were: CSSlow < 26.9, CSSmid 26.9 – 60.0, CSShigh > 60.0, with the highest CSS in our study being 409.7 (Table II). Although the ARTS-II investigators found the CSS to be superior to the SRS for predicting repeat revascularization and major adverse cardiovascular and cerebrovascular events (‘MACCE’) at 5 years after PCI, the fact that few very high risk patients were included in that study may have influenced their findings [21]. Importantly, the ARTS-II investigators found improved risk score accuracy when only patients with triple vessel CAD were considered [21], suggesting intrinsic patient risk influences risk score discriminatory ability. Consistent with this possibility, the ACEF score had an AUC for mortality of 0.65 in that study [21], which improved to 0.7586 in the current study. Therefore, these findings draw attention to the fact that, just as pretest probability is known to affect the accuracy of a given clinical test, angiographic features and clinical comorbidities may affect risk score accuracy. Moreover, this highlights the importance of properly validating any proposed risk score in the intended clinical patient population, rather than in carefully selected trial subjects. The importance of score calibration and validation in a relevant patient population has been previously suggested in PCI versus CABG patients, where it is known that while certain models may have good predictive ability in the setting of PCI, these models perform sub-optimally in similar patients undergoing CABG [24].
Complexity of predicting outcomes after PCI
Our study indicates that competing angiographic and clinical factors predict differing clinical outcomes after PCI for triple vessel CAD, and that there are inherent trade-offs made in risk prediction models when outcomes such as mortality, MI and TLR are combined as a composite MACE outcome. These conclusions are supported by Capodanno et al, as they identified that while the ACEF, SRS, CSS, Global Risk Classification (GRC) score and the EuroSCORE showed good predictive ability for cardiac mortality, all five scores showed reduced ability to predict MACE [24]. In addition, other groups have reported similar trends for the CSS [21] and SRS [25]. Therefore, our findings suggest that in clinical practice these differing outcomes may need to be considered in parallel, rather than in combination as MACE, and it is conceivable that more than one risk score may be required to completely evaluate a patient’s overall revascularization or PCI risk profile. Certainly, other specialized PCI risk scores have already been developed for predicting bleeding [31] and contrast nephropathy [32], while a new risk score derived using the National Cardiovascular Data Registry was recently described that includes only patients ≥ 65 years of age and which further segregates subjects based on presence of STEMI [33]. Furthermore and adding even greater complexity, another recent study based on post-CABG outcomes has suggested the predictive importance of differing factors entered into any risk model may be time-dependent, such that shock or emergency status are strong predictors of short-term outcomes, but for early survivors these factors are not predictive of long-term outcomes [34]. Our data add to these studies and strongly suggest that the very different outcomes of MI, mortality and TLR may require separate risk algorithms if it is deemed desirable to predict the likelihood of these events after PCI.
Yet another aspect to be considered in any PCI risk assessment is ‘bail-out’ status or hemodynamic instability. By their nature, these are high risk patients with an increased incidence of adverse events [19]. To avoid patient heterogeneity and address the most common clinical scenario, our study included only stable patients and we excluded those with STEMI presentation or shock. Importantly, 4 of the 6 scores we evaluated do not take any elements of bail-out or unstable patients into account (SRS, CSS, ACEF, ACEFmod), with only the NYSRS and NCDR scores having any domains that capture these aspects (Supplementary Table I). Of these, the NCDR score has by far the most comprehensive assessment of hemodynamic instability and bail-out status, and we speculate that our results may have differed if our study evaluated this alternate patient group. This underscores the fact that accurate PCI risk assessment is a surprisingly complex undertaking, and that very careful consideration needs to be given to applying an appropriate risk algorithm to any given patient.
Limitations
This was not a prospectively conducted study, but rather, the data of a single center registry were prospectively collected and subsequently analyzed. The population studied (n = 584) was modest in size, but remains larger than several recent comparable studies [21,22,25]. Nevertheless, the total mortality was only 40 patients, total number of MIs was 25, and total number of TLRs was 56 (Table IV). As a result, minor variation in assignation to tertiles for any given risk score may have influenced our results. Our relatively modest procedural success rate (97.4%) reflects the anatomic lesion complexity of this patient group. We were not able to include every possible risk score in this analysis, and in particular we did not consider the EuroSCORE [35], Global Risk Classification (GRC) score [22], the new NCDR score for patients aged ≥ 65 [33], or the new SYNTAX score II [8]. However, as it was not our aim to define the ‘best’ risk score, the omission of these and other scores is unlikely to have influenced our overall conclusions. Finally, not all risk scores were designed to evaluate the outcomes considered in this study. In particular, some scores were based on in-hospital data (e.g. NYSRS, NCDR, ACEF), while others used longer-term follow-up data (e.g. CSS). The SRS was not directly based on outcomes, but was created to evaluate anatomical complexity. Nevertheless, the comparisons made in this study are clinically relevant and consistent with the use of these scores in everyday practice.
Conclusions
Numerous risk scores now exist for predicting outcomes following PCI, and no score has clearly emerged as being superior. In our ‘all comers’ study of patients undergoing PCI for triple vessel and/or left main coronary disease, we compared six contemporary PCI risk scores and found that while several showed utility for predicting mortality, only the SRS and CSS were predictive of MI and no score was particularly effective for predicting TLR or overall MACE. Our data indicate that in everyday clinical practice, the factors that predispose to death may be very different from those that predispose to TLR, or to MI. Furthermore, these factors appear to partially nullify each other when overall MACE is considered. The present findings clarify the importance of these individual clinical outcomes and their respective predisposing factors, and suggest that physicians may need to tailor and individualize patient risk profile assessment prior to revascularization.
Supplementary Material
Acknowledgments
Funding sources: No specific funding or grant was used or acquired to conduct this study. Jason Kovacic is supported by National Institutes of Health Grant K08HL111330.
REFERENCES
- 1.Serruys PW, Onuma Y, Garg S, Sarno G, van den Brand M, Kappetein AP, Van Dyck N, Mack M, Holmes D, Feldman T, Morice MC, Colombo A, Bass E, Leadley K, Dawkins KD, van Es GA, Morel MA, Mohr FW. Assessment of the SYNTAX score in the Syntax study. EuroIntervention. 2009;5:50–56. doi: 10.4244/eijv5i1a9. [DOI] [PubMed] [Google Scholar]
- 2.Sianos G, Morel MA, Kappetein AP, Morice MC, Colombo A, Dawkins K, van den Brand M, Van Dyck N, Russell ME, Mohr FW, Serruys PW. The SYNTAX Score: an angiographic tool grading the complexity of coronary artery disease. EuroIntervention. 2005;1:219–227. [PubMed] [Google Scholar]
- 3.Serruys PW, Morice MC, Kappetein AP, Colombo A, Holmes DR, Mack MJ, Stahle E, Feldman TE, van den Brand M, Bass EJ, Van Dyck N, Leadley K, Dawkins KD, Mohr FW. Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease. N Engl J Med. 2009;360:961–972. doi: 10.1056/NEJMoa0804626. [DOI] [PubMed] [Google Scholar]
- 4.Wijns W, Kolh P, Danchin N, Di Mario C, Falk V, Folliguet T, Garg S, Huber K, James S, Knuuti J, Lopez-Sendon J, Marco J, Menicanti L, Ostojic M, Piepoli MF, Pirlet C, Pomar JL, Reifart N, Ribichini FL, Schalij MJ, Sergeant P, Serruys PW, Silber S, Sousa Uva M, Taggart D, Vahanian A, Auricchio A, Bax J, Ceconi C, Dean V, Filippatos G, Funck-Brentano C, Hobbs R, Kearney P, McDonagh T, Popescu BA, Reiner Z, Sechtem U, Sirnes PA, Tendera M, Vardas PE, Widimsky P, Alfieri O, Dunning J, Elia S, Kappetein P, Lockowandt U, Sarris G, Vouhe P, von Segesser L, Agewall S, Aladashvili A, Alexopoulos D, Antunes MJ, Atalar E, Brutel de la Riviere A, Doganov A, Eha J, Fajadet J, Ferreira R, Garot J, Halcox J, Hasin Y, Janssens S, Kervinen K, Laufer G, Legrand V, Nashef SA, Neumann FJ, Niemela K, Nihoyannopoulos P, Noc M, Piek JJ, Pirk J, Rozenman Y, Sabate M, Starc R, Thielmann M, Wheatley DJ, Windecker S, Zembala M. Guidelines on myocardial revascularization: The Task Force on Myocardial Revascularization of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS) Eur Heart J. 2010;31:2501–2555. doi: 10.1093/eurheartj/ehq277. [DOI] [PubMed] [Google Scholar]
- 5.Levine GN, Bates ER, Blankenship JC, Bailey SR, Bittl JA, Cercek B, Chambers CE, Ellis SG, Guyton RA, Hollenberg SM, Khot UN, Lange RA, Mauri L, Mehran R, Moussa ID, Mukherjee D, Nallamothu BK, Ting HH. 2011 ACCF/AHA/SCAI Guideline for Percutaneous Coronary Intervention: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines and the Society for Cardiovascular Angiography and Interventions. Circulation. 2011;124:2574–2609. doi: 10.1161/CIR.0b013e31823a5596. [DOI] [PubMed] [Google Scholar]
- 6.Serruys PW, Onuma Y, Garg S, Vranckx P, De Bruyne B, Morice MC, Colombo A, Macaya C, Richardt G, Fajadet J, Hamm C, Schuijer M, Rademaker T, Wittebols K, Stoll HP. 5-year clinical outcomes of the ARTS II (Arterial Revascularization Therapies Study II) of the sirolimus-eluting stent in the treatment of patients with multivessel de novo coronary artery lesions. J Am Coll Cardiol. 2010;55:1093–1101. doi: 10.1016/j.jacc.2009.11.049. [DOI] [PubMed] [Google Scholar]
- 7.Garg S, Sarno G, Girasis C, Vranckx P, de Vries T, Swart M, Bressers M, Garcia-Garcia HM, van Es GA, Raber L, Campo G, Valgimigli M, Dawkins KD, Windecker S, Serruys PW. A Patient-Level Pooled Analysis Assessing the Impact of the SYNTAX (Synergy Between Percutaneous Coronary Intervention With Taxus and Cardiac Surgery) Score on 1-Year Clinical Outcomes in 6,508 Patients Enrolled in Contemporary Coronary Stent Trials. JACC Cardiovasc Interv. 2011;4:645–653. doi: 10.1016/j.jcin.2011.02.018. [DOI] [PubMed] [Google Scholar]
- 8.Farooq V, van Klavern D, Steyerberg EW, Meliga E, Vergouwe Y, Chieffo A, Kappetein AP, Colombo A, Holmes DR, Mack M, Feldman T, Morice MC, Stahle E, Onuma Y, Morel MA, Garcia-Garcia HM, van Es GA, Dawkins KD, Mohr FW, Serruys PW. Anatomical and clinical characteristics to guide decision making between coronary artery bypass surgery and percutaneous coronary intervention for individual patients: development and validation of SYNTAX score II. Lancet. 2013;381:639–650. doi: 10.1016/S0140-6736(13)60108-7. [DOI] [PubMed] [Google Scholar]
- 9.Singh M, Lennon RJ, Holmes DR, Jr, Bell MR, Rihal CS. Correlates of procedural complications and a simple integer risk score for percutaneous coronary intervention. J Am Coll Cardiol. 2002;40:387–393. doi: 10.1016/s0735-1097(02)01980-0. [DOI] [PubMed] [Google Scholar]
- 10.Singh M, Rihal CS, Selzer F, Kip KE, Detre K, Holmes DR. Validation of Mayo Clinic risk adjustment model for in-hospital complications after percutaneous coronary interventions, using the National Heart, Lung, and Blood Institute dynamic registry. J Am Coll Cardiol. 2003;42:1722–1728. doi: 10.1016/j.jacc.2003.05.007. [DOI] [PubMed] [Google Scholar]
- 11.Singh M, Rihal CS, Lennon RJ, Spertus J, Rumsfeld JS, Holmes DR., Jr Bedside estimation of risk from percutaneous coronary intervention: the new Mayo Clinic risk scores. Mayo Clin Proc. 2007;82:701–708. doi: 10.4065/82.6.701. [DOI] [PubMed] [Google Scholar]
- 12.O’Connor GT, Malenka DJ, Quinton H, Robb JF, Kellett MA, Jr, Shubrooks S, Bradley WA, Hearne MJ, Watkins MW, Wennberg DE, Hettleman B, O’Rourke DJ, McGrath PD, Ryan T, Jr, VerLee P. Multivariate prediction of in-hospital mortality after percutaneous coronary interventions in 1994–1996. Northern New England Cardiovascular Disease Study Group. J Am Coll Cardiol. 1999;34:681–691. doi: 10.1016/s0735-1097(99)00267-3. [DOI] [PubMed] [Google Scholar]
- 13.Resnic FS, Ohno-Machado L, Selwyn A, Simon DI, Popma JJ. Simplified risk score models accurately predict the risk of major in-hospital complications following percutaneous coronary intervention. Am J Cardiol. 2001;88:5–9. doi: 10.1016/s0002-9149(01)01576-4. [DOI] [PubMed] [Google Scholar]
- 14.Qureshi MA, Safian RD, Grines CL, Goldstein JA, Westveer DC, Glazier S, Balasubramanian M, O’Neill WW. Simplified scoring system for predicting mortality after percutaneous coronary intervention. J Am Coll Cardiol. 2003;42:1890–1895. doi: 10.1016/j.jacc.2003.06.014. [DOI] [PubMed] [Google Scholar]
- 15.Block PC, Peterson ED, Krone R, Kesler K, Hannan E, O’Connor GT, Detre K. Identification of variables needed to risk adjust outcomes of coronary interventions: evidence-based guidelines for efficient data collection. J Am Coll Cardiol. 1998;32:275–282. doi: 10.1016/s0735-1097(98)00208-3. [DOI] [PubMed] [Google Scholar]
- 16.Shaw RE, Anderson HV, Brindis RG, Krone RJ, Klein LW, McKay CR, Block PC, Shaw LJ, Hewitt K, Weintraub WS. Development of a risk adjustment mortality model using the American College of Cardiology-National Cardiovascular Data Registry (ACC-NCDR) experience: 1998–2000. J Am Coll Cardiol. 2002;39:1104–1112. doi: 10.1016/s0735-1097(02)01731-x. [DOI] [PubMed] [Google Scholar]
- 17.Marks DS, Mensah GA, Kennard ED, Detre K, Holmes DR., Jr Race, baseline characteristics, and clinical outcomes after coronary intervention: The New Approaches in Coronary Interventions (NACI) registry. Am Heart J. 2000;140:162–169. doi: 10.1067/mhj.2000.106645. [DOI] [PubMed] [Google Scholar]
- 18.Wu C, Hannan EL, Walford G, Ambrose JA, Holmes DR, Jr, King SB, 3rd, Clark LT, Katz S, Sharma S, Jones RH. A risk score to predict in-hospital mortality for percutaneous coronary interventions. J Am Coll Cardiol. 2006;47:654–660. doi: 10.1016/j.jacc.2005.09.071. [DOI] [PubMed] [Google Scholar]
- 19.Peterson ED, Dai D, DeLong ER, Brennan JM, Singh M, Rao SV, Shaw RE, Roe MT, Ho KK, Klein LW, Krone RJ, Weintraub WS, Brindis RG, Rumsfeld JS, Spertus JA. Contemporary mortality risk prediction for percutaneous coronary intervention: results from 588,398 procedures in the National Cardiovascular Data Registry. J Am Coll Cardiol. 2010;55:1923–1932. doi: 10.1016/j.jacc.2010.02.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ranucci M, Castelvecchio S, Menicanti L, Frigiola A, Pelissero G. Risk of assessing mortality risk in elective cardiac operations: age, creatinine, ejection fraction, and the law of parsimony. Circulation. 2009;119:3053–3061. doi: 10.1161/CIRCULATIONAHA.108.842393. [DOI] [PubMed] [Google Scholar]
- 21.Garg S, Sarno G, Garcia-Garcia HM, Girasis C, Wykrzykowska J, Dawkins KD, Serruys PW. A new tool for the risk stratification of patients with complex coronary artery disease: the Clinical SYNTAX Score. Circ Cardiovasc Interv. 2010;3:317–326. doi: 10.1161/CIRCINTERVENTIONS.109.914051. [DOI] [PubMed] [Google Scholar]
- 22.Capodanno D, Miano M, Cincotta G, Caggegi A, Ruperto C, Bucalo R, Sanfilippo A, Capranzano P, Tamburino C. EuroSCORE refines the predictive ability of SYNTAX score in patients undergoing left main percutaneous coronary intervention. Am Heart J. 2010;159:103–109. doi: 10.1016/j.ahj.2009.10.021. [DOI] [PubMed] [Google Scholar]
- 23.Singh M, Rihal CS, Lennon RJ, Garratt KN, Holmes DR., Jr Comparison of Mayo Clinic risk score and American College of Cardiology/American Heart Association lesion classification in the prediction of adverse cardiovascular outcome following percutaneous coronary interventions. J Am Coll Cardiol. 2004;44:357–361. doi: 10.1016/j.jacc.2004.03.059. [DOI] [PubMed] [Google Scholar]
- 24.Capodanno D, Caggegi A, Miano M, Cincotta G, Dipasqua F, Giacchi G, Capranzano P, Ussia G, Di Salvo ME, La Manna A, Tamburino C. Global risk classification and clinical SYNTAX (synergy between percutaneous coronary intervention with TAXUS and cardiac surgery) score in patients undergoing percutaneous or surgical left main revascularization. JACC Cardiovasc Interv. 2011;4:287–297. doi: 10.1016/j.jcin.2010.10.013. [DOI] [PubMed] [Google Scholar]
- 25.Capodanno D, Di Salvo ME, Cincotta G, Miano M, Tamburino C. Usefulness of the SYNTAX score for predicting clinical outcome after percutaneous coronary intervention of unprotected left main coronary artery disease. Circ Cardiovasc Interv. 2009;2:302–308. doi: 10.1161/CIRCINTERVENTIONS.108.847137. [DOI] [PubMed] [Google Scholar]
- 26.Bamford JM, Sandercock PA, Warlow CP, Slattery J. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1989;20:828. doi: 10.1161/01.str.20.6.828. [DOI] [PubMed] [Google Scholar]
- 27.D’Agostino RB, Belanger A, D’Agostino RB. A Suggestion for Using Powerful and Informative Tests of Normality. The American Statistician. 1990;44:316–321. [Google Scholar]
- 28.DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44:837–845. [PubMed] [Google Scholar]
- 29.Kovacic JC, Fuster V. From Treating Complex Coronary Artery Disease to Promoting Cardiovascular Health: Therapeutic Transitions and Challenges, 2010–2020. Clin Pharmacol Ther. 2011;90:509–518. doi: 10.1038/clpt.2011.173. [DOI] [PubMed] [Google Scholar]
- 30.Roger VL, Go AS, Lloyd-Jones DM, Benjamin EJ, Berry JD, Borden WB, Bravata DM, Dai S, Ford ES, Fox CS, Fullerton HJ, Gillespie C, Hailpern SM, Heit JA, Howard VJ, Kissela BM, Kittner SJ, Lackland DT, Lichtman JH, Lisabeth LD, Makuc DM, Marcus GM, Marelli A, Matchar DB, Moy CS, Mozaffarian D, Mussolino ME, Nichol G, Paynter NP, Soliman EZ, Sorlie PD, Sotoodehnia N, Turan TN, Virani SS, Wong ND, Woo D, Turner MB. Heart Disease and Stroke Statistics--2012 Update: A Report From the American Heart Association. Circulation. 2012;125:e2–e220. doi: 10.1161/CIR.0b013e31823ac046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Mehran R, Pocock SJ, Nikolsky E, Clayton T, Dangas GD, Kirtane AJ, Parise H, Fahy M, Manoukian SV, Feit F, Ohman ME, Witzenbichler B, Guagliumi G, Lansky AJ, Stone GW. A risk score to predict bleeding in patients with acute coronary syndromes. J Am Coll Cardiol. 2010;55:2556–2566. doi: 10.1016/j.jacc.2009.09.076. [DOI] [PubMed] [Google Scholar]
- 32.Mehran R, Aymong ED, Nikolsky E, Lasic Z, Iakovou I, Fahy M, Mintz GS, Lansky AJ, Moses JW, Stone GW, Leon MB, Dangas G. A simple risk score for prediction of contrast-induced nephropathy after percutaneous coronary intervention: development and initial validation. J Am Coll Cardiol. 2004;44:1393–1399. doi: 10.1016/j.jacc.2004.06.068. [DOI] [PubMed] [Google Scholar]
- 33.Weintraub WS, Grau-Sepulveda MV, Weiss JM, Delong ER, Peterson ED, O’Brien SM, Kolm P, Klein LW, Shaw RE, McKay C, Ritzenthaler LL, Popma JJ, Messenger J, Shahian DM, Grover FL, Mayer JE, Garratt K, Moussa I, Edwards FH, Dangas GD. Prediction of Long Term Mortality after Percutaneous Coronary Intervention in Older Adults: Results from the National Cardiovascular Data Registry. Circulation. 2012;125:1501–1510. doi: 10.1161/CIRCULATIONAHA.111.066969. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Shahian DM, O’Brien SM, Sheng S, Grover FL, Mayer JE, Jacobs JP, Weiss JM, Delong ER, Peterson ED, Weintraub WS, Grau-Sepulveda MV, Klein LW, Shaw RE, Garratt K, Moussa I, Shewan CM, Dangas GD, Edwards FH. Predictors of Long-Term Survival Following Coronary Artery Bypass Grafting Surgery: Results from The Society of Thoracic Surgeons Adult Cardiac Surgery Database (The ASCERT Study) Circulation. 2012;125:1491–1590. doi: 10.1161/CIRCULATIONAHA.111.066902. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Romagnoli E, Burzotta F, Trani C, Siviglia M, Biondi-Zoccai GG, Niccoli G, Leone AM, Porto I, Mazzari MA, Mongiardo R, Rebuzzi AG, Schiavoni G, Crea F. EuroSCORE as predictor of in-hospital mortality after percutaneous coronary intervention. Heart. 2009;95:43–48. doi: 10.1136/hrt.2007.134114. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.