High-risk percutaneous coronary intervention (PCI) procedures are being performed in greater numbers, in older patients with multiple comorbidities, and increasingly in the setting of acute coronary syndromes. Estimating inpatient PCI mortality and MACE risk (mortality, Q-wave myocardial infarction, urgent coronary artery bypass grafting and stroke) is essential in informing decision-making, consent, and operator and institutional benchmarking. There are a number of currently available risk scores that are often applied interchangeably. We investigated if there was concordance between contemporary risk scoring systems for inpatient mortality and MACE following PCI in patients at low, moderate or high-risk in a ‘real life’ cohort, depending upon method of presentation (elective, urgent, emergency).Methods
We retrospectively identified 1,404 consecutive patients treated by PCI within a 6-month period in 2013. The New York risk score (NY) and National Cardiovascular Data Registry score (NCDR) were calculated for each patient to predict inpatient mortality risk and the Northwestern Quality Improvement score (NWQIP) and the Mayo Clinical Risk Score (MCRS) were calculated to predict inpatient MACE risk. Using the NY score as the reference for inpatient mortality and the NWQIP score as the reference for inpatient MACE, patients were divided into three risk groups (low < 1%; moderate 1–5%; high > 5%) and stratified into elective, urgent and emergency clinical presentations. Concordance was estimated using the Intraclass Correlation Coefficient (ICC) calculated with respect to each risk score and stratified by patient group.Results
757 patients were identified as low-risk (461 elective, 280 urgent, 16 emergency), 497 patients as moderate-risk (73 elective, 197 urgent, 227 emergency) and 150 patients as high-risk (4 elective, 43 urgent, 103 emergency) for inpatient mortality. 607 patients were identified as low-risk (382 elective, 225 urgent, 0 emergency), 594 patients as moderate-risk (97 elective, 237 urgent, 260 emergency) and 203 patients as high-risk (14 elective, 66 urgent, 123 emergency) for inpatient MACE events. The ICC indicated that risk scores correlated well for low-risk groups however were poorly concordant for high-risk groups (Table 1) and that this was true regardless of mode of clinical presentation (Table 2).Conclusions
Currently available PCI risk scores for both inpatient mortality and MACE are broadly concordant in low risk patient groups. However in patients at higher risk with multiple co-morbid factors, current tools are poorly concordant in predicting mortality and MACE. This has important implications for consent, and benchmark analysis of operator and institutional performance. Better understanding of choice of risk score for use in clinical practice is needed particularly as case mix and complexity evolve.