To describe the incidence, predictors, and in-hospital impact of early SVA (eSVA, occurring<day 2 post-STEMI) and late SVA (lSVA, occurring≥day 2 post-STEMI) in STEMI patients. To derive and internally validate a risk score to identify patients at high-risk of lSVA.

Data of 13523 patients enrolled in the ORBI registry were analysed. Logistic regression was performed to identify predictors of eSVA, lSVA, and in-hospital all-cause mortality. Predictors of lSVA were used to build a risk score.

eSVA occurred in 678 patients (5%) whereas 120 patients (0.9%) experienced lSVA at a median timing of 3 days post-STEMI. eSVA associated with a significantly higher risk of all-cause mortality (adjusted OR: 1.90, 95%CI: 1.39–2.61, P<0.001) whereas only a trend was observed with lSVA (adjusted OR: 1.69, 95%CI: 0.91–3.13, P=0.09). Multivariable predictors of eSVA are listed in Table 1. Multivariable predictors of lSVA are listed in the Figure. The score derived from these variables allowed the classification of patients into four risk categories: low (0–21), low-to-intermediate (22–34), intermediate-to-high (35–44), and high (≥45). Observed lSVA rates were 0.2%, 0.4%, 0.8%, and 2.5%, across the four risk categories, respectively. The model demonstrated good discrimination (20-fold cross-validated c-statistic of 0.76) and adequate calibration (Hosmer-Lemeshow P=0.40).

eSVA are 5-fold more common than lSVA in the setting of STEMI, mainly associated with other early complications, and portends a 2-fold higher risk of in-hospital mortality. Moreover, we developed a risk score-identifying patients at high risk of lSVA for whom early ICU discharge may not be suitable.