When sPESI falls short: the msPESI raises the bar in pulmonary embolism risk stratification

Background

Pulmonary embolism (PE) remains a significant cause of in-hospital mortality in Intensive Care Units (ICUs). Existing prognostic models often fail to accurately identify patients at highest risk who may benefit from more intensive management.

Aim

To identify independent predictors of poor prognosis and to develop a modified predictive score for patients with intermediate-high and high-risk PE.

Methods

This retrospective, observational, single-center study included patients with intermediate-high or high-risk PE (according to the ESC guidelines1) admitted to the ICU between January 2020 and December 2024. Independent predictors of outcomes were identified using Multivariate Cox regression analysis. The simplified Pulmonary Embolism Severity Index (sPESI) was modified to create a new model – the modified sPESI (msPESI) – which was then compared with the original sPESI for prognostic performance.

Results

A total of 87 patients were included (52% of female sex, with a mean age of 56±17 years). Of these, 8% had previous PE, 9% had history of venous thromboembolism, 21% had history of cancer, 16% had thrombophilia and 30% had recent bed rest. On admission, mean hemoglobin (Hb) was 12.1±2.9 g/dL, median T troponin was 85 [170] ng/L, median NTproBNP was 2560 [5508] pg/dL and median lactate was 22.0 [38.8] mg/dL. Only 1 patient had sPESI=0 and the remaining had sPESI≥1. Regarding reperfusion strategies, 43% of the patients received fibrinolysis and 5% underwent thrombectomy. During ICU stay, 63% of the patients developed obstructive shock, 33% had cardiac arrest, 9% required extracorporeal membrane oxygenation, 25% experienced major bleeding and 32% died. Higher sPESI scores were independently associated with 30-day mortality (HR 1.774, 95% CI 1.079-2.918, p=0.024). In multivariate Cox regression (adjusted for age, sex, NT-proBNP and troponin), independent predictors of events were lactate (HR 1.014, 95%, CI 1.006-1.022, p<0.001) and hemoglobin at admission (HR 0.762, 95%, CI 0.649-0.898, p=0.001). Hence, we developed the modified score, using variables readily obtainable from arterial blood gas analysis: msPESI = 0.573 * sPESI + 0.014 * Lactate (mg/dL) – 0.272 * [Hb (g/dL) – 13] + 0.35.

The msPESI outperformed sPESI in predicting 30-day mortality (AUC 0.854 vs. 0.741) (Figure 1). The optimal cut-off value for predicting 30-day mortality was msPESI=2.5, yielding 89% sensitivity and 66% specificity. Patients with an msPESI>2.5 had a 7.6-fold higher risk of 30-day mortality (HR 7.651, 95% CI 2.639-22.185, p<0.001) (Figure 2). Higher msPESI scores were strongly associated with increased 30-day mortality (HR 2.298, 95% CI 1.752-3.016, p<0.001) (Figure 3).

Conclusions

The msPESI outperformed the original index in predicting short-term mortality among critically ill patients with PE. It is a simple tool, which integrates two easily obtainable parameters, which may guide therapeutic strategies.For image description, please refer to the figure legend and surrounding text.For image description, please refer to the figure legend and surrounding text.