Validation of the Mayo Clinic Risk Score for In-Hospital Mortality After Percutaneous Coronary Interventions Using the National Cardiovascular Data Registry

Machine learning prediction of one-year mortality after percutaneous coronary intervention in acute coronary syndrome patients

BACKGROUND

Machine learning (ML) models have the potential to accurately predict outcomes and offer novel insights into inter-variable correlations. In this study, we aimed to design ML models for the prediction of 1-year mortality after percutaneous coronary intervention (PCI) in patients with acute coronary syndrome.

METHODS

This study was performed on 13,682 patients at Tehran Heart Center from 2015 to 2021. Patients were split into 70:30 for testing and training. Four ML models were designed: a traditional Logistic Regression (LR) model, Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Ada Boost models. The importance of features was calculated using the RF feature selector and SHAP based on the XGBoost model. The Area Under the Receiver Operating Characteristic Curve (AUC-ROC) for the prediction on the testing dataset was the main measure of the model's performance.

RESULTS

From a total of 9,073 patients with >1-year follow-up, 340 participants died. Higher age and higher rates of comorbidities were observed in these patients. Body mass index and lipid profile demonstrated a U-shaped correlation with the outcome. Among the models, RF had the best discrimination (AUC 0.866), while the highest sensitivity (80.9%) and specificity (88.3%) were for LR and XGBoost models, respectively. All models had AUCs of >0.8.

CONCLUSION

ML models can predict 1-year mortality after PCI with high performance. A classic LR statistical approach showed comparable results with other ML models. The individual-level assessment of inter-variable correlations provided new insights into the non-linear contribution of risk factors to post-PCI mortality.

Comparison of SYNTAX score strata effects of percutaneous and surgical revascularization trials: A meta-analysis

OBJECTIVES

The evidence supporting the use of the Synergy Between Percutaneous Coronary Intervention With TAXUS and Cardiac Surgery (SYNTAX) score for risk stratification is controversial. We performed a systematic review and meta-analysis of all the randomized controlled trials comparing percutaneous coronary intervention versus coronary artery bypass grafting that reported their outcomes stratified by SYNTAX score, focusing on between-strata comparisons.

METHODS

A systematic review of MEDLINE, EMBASE, Cochrane Library databases was performed. Incidence rate ratios were pooled with a random effect model. Between-group statistical heterogeneity according to accepted SYNTAX score tertiles was computed in the main analysis. Ratios of incidence rate ratios were computed to appraise between-strata effect, as sensitivity analysis. Primary and secondary outcomes were major adverse cardiac and cerebrovascular events and all-cause mortality, respectively. Separate sub-analyses were performed for left main and multivessel disease.

RESULTS

From 425 citations, 6 trials were eventually included (8269 patients [4134 percutaneous coronary interventions, 4135 coronary artery bypass graftings]; mean follow-up: 6.2 years [range: 3.8-10]). Overall, percutaneous coronary intervention was associated with a significant increase in major adverse cardiac and cerebrovascular events (incidence rate ratio, 1.39, 95% confidence interval, 1.27-1.51) and nonsignificant increase in all-cause mortality (incidence rate ratio, 1.17, 95% confidence interval, 0.98-1.40). There was no significant statistical heterogeneity of treatment effect by SYNTAX score for major adverse cardiac and cerebrovascular events or mortality (P = .40 and P = .34, respectively). Results were consistent also for patients with left main and multivessel disease (major adverse cardiac and cerebrovascular events: P = .85 in left main, P = .78 in multivessel disease 0.78; mortality: P = .12 in left main; P = .34 in multivessel disease). Results of analysis based on ratios of incidence rate ratios were consistent with the main analysis.

CONCLUSIONS

No significant association was found between SYNTAX score and the comparative effectiveness of percutaneous coronary intervention and coronary artery bypass grafting. These findings have implications for clinical practice, future guidelines, and the design of percutaneous coronary intervention versus coronary artery bypass grafting trials.

Derivation and external validation of a novel risk score for prediction of 30-day mortality after percutaneous coronary intervention

AIMS

Early mortality after percutaneous coronary intervention (PCI) is relatively rare. Current risk prediction models for this event are outdated. We sought to derive a 30-day mortality risk score after PCI.

METHODS AND RESULTS

The score was derived from a pooled database of 21 randomised clinical trials using a logistic regression model incorporating clinical and angiographic variables. The score was validated in a separate unrestricted study population, the Assessment of Dual AntiPlatelet Therapy With Drug Eluting Stents (ADAPT-DES) registry. Of 32,882 eligible patients, 75% had data for all 19 variables used for score derivation. The independent predictors of 30-day mortality were age, presentation with ACS, diabetes mellitus, use of first-generation drug-eluting stents, left main or left anterior descending artery lesion, prior myocardial infarction (MI), and suboptimal flow in the artery before or after PCI. The median [interquartile range] score in the derivation cohort was 5 [3, 6] and overall mortality was 0.49%, ranging from 0.08% to 1.64% with scores of 0-16. The 30-day mortality rate was approximately tenfold higher in patients with a score at or above versus below the median of 5 (0.86% versus 0.08%, p<0.0001). Discrimination in both cohorts was very good (C statistic=0.848 and 0.828, respectively), and calibration was satisfactory.

CONCLUSIONS

A novel risk score incorporating eight readily available clinical and angiographic variables had high discrimination for 30-day death after PCI across a wide range of clinical scenarios.

Subsample analysis of the Vascular Complications Risk Score at two public referral centers for interventional cardiology

OBJECTIVE

Evaluate the performance of the Vascular Complications Risk Score in two public referral centers for interventional cardiology.

METHOD

Subsample analysis of the Vascular Complications Risk Score, which was developed and validated in the catheterization laboratories of three cardiology referral centers (two public, one private) with a cutoff of <3 for no risk of developing vascular complications and ≥3 for risk. In this new analysis, we excluded data from the private facility, and only included participants from the original (validation) cohort of the two public hospitals.

RESULTS

Among the 629 patients studied, 11.8% had vascular complications; of these, 1.8% were major and 10% minor. Among the patients with a score <3, 310 (94.5%) presented no vascular complications; of those with a score ≥3, 50 (17%) developed complications. Of those who developed vascular complications, 18 scored <3; two of these had major complications.

CONCLUSION

This subanalysis confirms the ability of the Vascular Complications Risk core to predict low risk of vascular complications in patients with a score < 3.

Vascular complications in patients who underwent endovascular cardiac procedures: multicenter cohort study

OBJECTIVE

to analyze vascular complications among patients who underwent endovascular cardiac procedures in the hemodynamic laboratories of three referral centers.

METHOD

a multicenter cohort study was conducted in three referral facilities. The sample was composed of 2,696 adult patients who had undergone elective or urgent percutaneous cardiac procedures. The outcomes were vascular complications, such as: hematoma at the site of the arterial puncture; major or minor bleeding; surgical correction for retroperitoneal hemorrhage; pseudoaneurysm; and arteriovenous fistula.

RESULTS

237 (8.8%) of the 2,696 patients presented a vascular complication at the site of the arterial puncture. The total number of vascular complications was 264: minor hematoma<10cm (n=135); stable bleeding (n=86); major hematoma ≥10cm (n=32); and unstable bleeding (n=11). There were no retroperitoneal hematoma events, pseudoaneurysm or arterial venous fistula. Most of the major and minor complications occurred in the first six hours after the procedure.

CONCLUSION

the results concerning the current context of interventional cardiology indicate that the complications predominantly occur in the first six hours after the procedure, considering a 48-hour follow-up. The staff should plan and implement preventive measures immediately after the procedures.

Clinical Outcomes of Polytetrafluoroethylene-Covered Stents for Coronary Artery Perforation in Elderly Patients Undergoing Percutaneous Coronary Interventions

Background

Coronary artery perforation (CAP) during percutaneous coronary intervention (PCI) is associated with increased mortality. Polytetrafluoroethylene covered stents (CS) are an effective approach to treat CAP, but data regarding elderly patients requiring CS implantation for CAP are limited. The aim of this study is to report clinical data for elderly CAP patients undergoing CS implantation during PCI.

Methods

Nineteen consecutive elderly patients (≥ 65 years) undergoing CS implantation due to PCI-induced CAP in a tertiary referral center from July 2003 to April 2016 were retrospectively examined.

Results

There were 13 men and six women, with a mean age of 75.3 ± 5.6 years (range: 65-86 years). Perforation grade was Ellis type II in five patients (26.3%), and Ellis type III in 14 patients (73.7%). Cardiac tamponade developed in six patients (31.6%), and intra-aortic balloon pumping was needed in four patients (21.1%). The overall success rate for CS implantation rate was 94.7%. The overall in-hospital mortality rate was 15.8%; the in-hospital myocardial infarction rate was 63.2%. Among 16 survival-to-discharge cases, dual antiplatelet therapy (DAPT) was prescribed in 14 cases (87.5%) for a mean duration of 14 months. Overall, there were five angiogram- proven CS failures among 18 patients receiving successful CS implantation. The 1, 2 and 4 years of actuarial freedom from the CS failure were 78%, 65%, and 43% in the angiogram follow-up patients.

Conclusions

CS implantation for CAP is feasible and effective in elderly patients, while CS failure remains a major concern that encourages regular angiographic follow-up in these case.

Assessment of individual operator performance using a risk-adjustment model for percutaneous coronary interventions

OBJECTIVE

To investigate the applicability of the Mayo Clinic Risk Score (MCRS) in the assessment of performance metrics of individual interventional cardiologists at 3 Mayo Clinic sites.

PARTICIPANTS AND METHODS

We evaluated the risk-adjusted performance of 21 interventional cardiologists who performed 8187 percutaneous coronary intervention procedures at 3 Mayo Clinic sites from January 1, 2007, through December 31, 2010. Observed mortality, major adverse cardiac events (MACEs) (eg, death, Q-wave myocardial infarction, urgent or emergent coronary artery bypass graft, and stroke), and expected risk were estimated using the MCRS. To compare individual performance against the other operators, risk estimates were recalibrated by excluding the individual performer from logistic regression models.

RESULTS

The log odds ratio for observed vs estimated risk was estimated for each interventional cardiologist, and their individual effects were then plotted on a normal probability plot to identify outliers. Observed in-hospital mortality was not different than expected (1.8% vs 1.6%; P=.24); however, the postprocedural MACE rate was lower than predicted (observed, 2.7%; expected, 3.8%; P<.001). All but one interventional cardiologist had MACE and death rates within the expected variation. Detailed assessment of that operator's risk performance produced excellent outcomes (observed vs expected MACE rate, 1.0% vs 4.4%).

CONCLUSION

The MCRS can serve as a tool for the assessment of performance metrics for interventional cardiologists, and risk-adjusted outcomes may serve as a better surrogate for institutional quality metrics.

Risk scores in acute coronary syndrome and percutaneous coronary intervention: a review

Patients with acute coronary syndrome (ACS) need to be risk stratified to deliver the most appropriate therapy. The GRACE and TIMI risk scores have penetrated contemporary guidelines with the former most commonly used in clinical practice. However, ACS prediction models need to be re-evaluated in contemporary practice with evolving diagnostic and treatment options. Moreover, the increased availability of percutaneous coronary intervention (PCI) as a treatment option for ACS combined with an expanding case mix and emphasis on quality control have triggered the creation of PCI specific prognostic models. These allow clinicians and patients to have an understanding of expected outcomes following PCI by predicting outcomes in-hospital to 5 years following intervention. The aim of this review is to evaluate the most recognized and studied ACS/PCI risk models, focusing on their strengths and limitations, and to assess the need for more robust tools to predict outcomes in a period of constantly advancing technologies and changing patient demographics.

The impact of achieving perfect care in acute coronary syndrome: the role of computer assisted decision support

BACKGROUND

Attainment of every performance measure or perfect care (PC) is used as a tool for measuring hospital quality of care. We sought to describe the effect of achieving PC on subsequent outcomes in patients admitted with acute coronary syndrome (ACS) and to determine whether computerized physician order entry enabled with decision support (CPOE-DS) would enhance the likelihood of achieving PC and improvements in clinical outcomes.

METHODS

Clinical inpatient data, performance measures and subsequent mortality was collected in 1,321 consecutive ACS patients admitted between January 1, 2009, to October 15, 2011, using either a standardized order set that followed consensus guidelines or orders generated via CPOE-DS.

RESULTS

CPOE-DS generated orders were utilized in 642 (49%) patients while the remaining 679 (51%) of patients were admitted using standardized order sets. At baseline, CPOE-DS patients were younger (-3%, P = .006), had lower resting heart rates (-3%, P = .012), higher TIMI risk scores (+19%, P < .001), were less likely to have hypertension (85% vs. 90%, P = .014), and more likely to have ST-segment elevation myocardial infarction (17% vs 10%; P = .001) than patients admitted with standard orders. Patients admitted using CPOE-DS were 5.7 times more likely to achieve PC than those who were admitted with standard orders (P < .001). Independent predictors of survival included PC (HR, 0.45; P < .001), age ≥67 years (HR, 2.34; P < .001), and abnormal presenting heart rate (HR, 1.71; P = .046).

CONCLUSIONS

Achievement of PC is a valid measure of quality of care in the hospitalized ACS patient and is associated with improved survival. CPOE-DS is feasible in the care process for ACS and can increase attainment of PC.

Which antithrombin for whom? Identifying the patient population that benefits most from novel antithrombin agents

Anticoagulation has proven to be a key component in the management of acute coronary syndromes (ACS). Pharmacological agents with various modes of action are utilized to reduce thrombus development by impairing thrombin formation, platelet activation, and platelet aggregation. The optimal management of these patients is to achieve maximal anti-ischemic benefit while avoiding bleeding complications. Synthetic "novel" agents have been developed to specifically target factor Xa or thrombin to achieve this goal. A growing amount of data show that these agents provide a net clinical benefit in the setting of stable ischemic heart disease, unstable angina, non-ST-elevation myocardial infarction (NSTEMI), and ST-elevation myocardial infarction (STEMI).