Adding frailty and post-PPCI LVEF<40% to the Zwolle Risk Score improved the prediction of in-hospital complications in STEMI patients (AUC increased from 0.693 to 0.804; p<0.001).
Observational (n=375)
No
Does adding frailty and post-PPCI LVEF to the Zwolle Risk Score improve the prediction of in-hospital complications in STEMI patients treated with primary PCI?
Adding early assessment of frailty and post-PPCI LVEF to the Zwolle Risk Score significantly improves the identification of STEMI patients at very low risk of in-hospital complications.
Effect estimate: AUC 0.804 (95% CI 0.716-0.892)
p-value: p=<0.001
Abstract Introduction Despite advances in the management of STEMI, the incidence of complications during hospitalisation is not exceptional. The identification of patients with a low risk of unfavourable short-term outcomes, who may need less surveillance and could be safely early discharged, is a matter of ongoing interest. In this regard, predictive models have been developed such as the Zwolle Risk Score (ZRS), which assigns various scores to 6 variables: age, Killip class, anterior location, three-vessel disease, TIMI flow post primary percutaneous coronary intervention (PPCI) and ischaemia time. A ZRS ≤3 appears to identify low-risk patients after successful PPCI. Purpose To evaluate the performance of the ZRS in STEMI patients treated with PPCI in our setting and to analyse whether clinical variables with potential predictive value recorded immediately after PPCI, such as frailty and left ventricular ejection fraction (LVEF), add relevant information to predict in-hospital complications. Methods Single-centre, prospective, observational study including consecutive patients with STEMI and reperfusion strategy with PPCI. Clinical and angiographic variables were recorded, including ZRS, Clinical Frailty Scale (CFS) and echocardiographic-estimated LVEF, obtained immediately after PPCI in the catheterisation laboratory. Their association with major complications defined as in-hospital mortality or major adverse cardiovascular events (MACE: cardiovascular death, non-lethal infarction or non-lethal stroke) was analysed by univariate and multivariate binary logistic regression, and their discriminatory capacity added to the ZRS using the area under the operator receiver curve (AUC). Results Out of 375 patients included, 220 (58.7%) were identified as low risk with ZRS≤3 presenting a mortality of 1.8% and a MACE rate of 4.1%, while in the ZRS3 group these were 18.2% and 22.9%, respectively. In the univariate analysis, we found an association with in-hospital complications and ZRS3, four of the six variables included in the ZRS, frailty (CFS≥4) and post-PPCI LVEF40%. We found no association between age60 years and ischaemia time (variables included in ZRS) and in-hospital complications. In multivariate analysis ZRS3, frailty (CFS≥4) and post-PPCI LVEF40% were associated with in-hospital complications (see Figure 1). The AUC of the variable ZRS3 for the prediction of in-hospital complications was 0.693 (95%CI 0.600-0.787; p0.001) and it raised to 0.804 (95%CI 0.716-0.892; p0.001) for the extended model that included the variables frailty and post-PPCI LVEF40% (see Figure 2). Conclusions In our setting, although patients with ZRS≤3 usually have a favourable hospital outcome, they have a non-negligible rate of complications. Early assessment of variables such as frailty and post-PPCI LVEF at the catheterisation lab may help to improve the identification of patients with very low risk of complications during hospitalisation.Univariate and multivariate analysis ROC curve for major complications
Santiago et al. (Sat,) conducted a observational in STEMI (n=375). Extended risk model (Zwolle Risk Score + frailty + post-PPCI LVEF) vs. Zwolle Risk Score alone was evaluated on In-hospital complications (mortality or MACE) (AUC 0.804, 95% CI 0.716-0.892, p=<0.001). Adding frailty and post-PPCI LVEF<40% to the Zwolle Risk Score improved the prediction of in-hospital complications in STEMI patients (AUC increased from 0.693 to 0.804; p<0.001).